Detecting Weligama Coconut Leaf Wilt Disease in Coconut Using UAV-Based Multispectral Imaging and Object-Based Classification

Detecting Weligama Coconut Leaf Wilt Disease in Coconut Using UAV-Based Multispectral Imaging and Object-Based Classification | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Detecting Weligama Coconut Leaf Wilt Disease in Coconut Using UAV-Based Multispectral Imaging and Object-Based Classification H.D.M.U Wijesinghe1 H.D.M.U Wijesinghe1, KMC Tahrupath, JAYASINGHE GUTTILA This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5767642/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 12 Jul, 2025 Read the published version in Journal of Plant Diseases and Protection → Version 1 posted 5 You are reading this latest preprint version Abstract Weligama Coconut Leaf Wilt Disease (WCLWD), a major threat to the coconut industry in Sri Lanka, has resulted in large economic losses (reduced productivity and high mortality rate) among infected palm. Early diagnosis is challenging and unreliable due to the low sensitivity of conventional disease detection methods like visual inspections and laboratory testing. In order to overcome these constraints, this study used object-based image analysis (OBIA) in combined with multispectral imaging using an unmanned aerial vehicle (UAV) to identify and categorize WCLWD in coconut palms. To differentiate between healthy and infected trees, Support Vector Machine (SVM) classification was used to analyze UAV images taken in five spectral bands: red, green, blue, red edge, and near infrared. The four band combination of 'blue', 'green', 'red-edge' and 'near infrared' was found to be the best of those tested, with an accuracy of 79.25% and a moderate agreement, based on the kappa coefficient of 0.493. The accuracy of this was then validated against a field survey ground truth data. Results show that overland biomass detection using OBIA methods with UAV multispectral imaging offers a feasible means to identify WCLWD, but that further classifier work and extra sources of data can improve accuracy. Results show the possibility of advanced remote sensing technologies for improve the detection of coconut WCLWD and support for managing the spread of disease in coconut plantations. Agricultural disease monitoring Coconut disease detection Object-based image analysis (OBIA) Remote sensing Unmanned aerial vehicle (UAV) Weligama Coconut Leaf Wilt Disease (WCLWD) Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Introduction Coconut cultivation, particularly in tropical regions, is economically and culturally significant due to its diverse uses, including food, oil, fiber, and medicine. The coconut palm ( Cocos nucifera ), occasionally called the tree of life, stands out as one of the most versatile trees globally. Coconut farming is one of the most important crops in the agricultural industry all over the world, mainly in Indonesia, the Philippines and India. The coconut market around the world is becoming wider due to people’s preference for foods and products made from coconut oil, water, and more that are considered healthy products in the food chain (Prades et al. 2016 ). Apart from being a source of income to millions of farmers, it contributes significantly to the economies of producing countries through exports and job creation (Prades et al. 2016 ). In Sri Lankan agricultural economy the most important tree crop is coconut., covering approximately 455,000 hectares as of 2015 and producing around 3,056 million nuts annually (Subajiny et al. 2018 ). Also contributes to about 0.5% of the GDP of the country and directly and indirectly uses the 1.5 million people (Perera et al. 2016a ). Sri Lanka has a very good market for copra and coconut oil that are highly demanded for use locally and in export markets. The genetic stock of coconuts in the country is diverse with the Sri Lanka Tall (Typica) as the most common progeny (Perera et al. 2016a ). Also, the sector has some challenges that affect productivity and sustainability such as climate change, pests, and diseases among them (Rajan 2011 ). Weligama Coconut Leaf Wilt Disease (WCLWD) is a severe threat, recognized in 2006 in the region of Weligama in Southern Sri Lanka and mainly on coconut trees (Nainanayake et al. 2016b ; Kanatiwela-de Silva et al. 2019 ). The causal agent is a phytoplasma. The disease is characterized by flaccidity and marginal necrosis of leaflets, intense yellowing of the fronds, reduction of the crown size and tapering of the trunk, loss of productivity and eventual death of the palm within two years (Perera et al. 2016b ). Figures 1 and 2 show symptoms of WCLWD (Wijesekara et al. 2008 ). The traditional methods of WCLWD detection are based on visual inspection, laboratory testing, and ground surveys. And among these, the most commonly used classical techniques include the nested PCR that makes it possible to quickly identify the phytoplasma causing WCLWD (Kanatiwela-de Silva et al. 2019 ). Also demonstrated that an indirect ELISA, validated alongside PCR, achieved high accuracy (93%) and sensitivity (92.7%) for detecting WCLWD-associated phytoplasma, although specificity was lower at 79% (Kanatiwela-de Silva et al. 2019 ). The nested PCR approach has also shown promise, with a success rate of 88% and 100% specificity in detecting phytoplasmas in coconut tissues (De Silva et al. 2023 ). Researchers have connected the phytoplasma to several environmental stresses that potentially worsen the disease’s effects (Perera et al. 2016a ). Uncertainty of early diagnosis of WCLWD is attributed to low titre of phytoplasma in the affected tissues (Wijesekara et al. 2020 ). Elimination of phytoplasma through organic management is not possible because there is no effective remedial measure to control phytoplasma other than developing resistant coconut cultivars (Perera et al. 2016a ). Disease spread has led to quarantining, as well as the uprooting of palms as a means of preventing the further spread of the disease (Nainanayake et al. 2016a , b). Information on the molecular characteristics is very useful to determine how the associated phytoplasma is best detected and managed (Kanatiwela-de Silva et al. 2019 ). But conventional methods like field visits have been used for assessing the incidences of the disease across district levels; study done in 2012, showed that 65,838, 251,980, and 14,344 palms were affected in Galle, Matara, and Hambantota, respectively (Nainanayake et al. 2016b ). However, these methods have their drawbacks; for example, it has been realized that spectral methods such as spectral analysis have low sensitivity in detecting palm areas affected by WCLWD (Nainanayake et al. 2016a ). Multispectral imaging has emerged as a pivotal technology in agricultural disease detection, particularly through the utilization of unmanned aerial vehicles (UAVs). This imaging technique captures data across multiple wavelengths, enabling the assessment of plant health by analyzing spectral signatures associated with various physiological conditions. Barman et al. ( 2023 ) employed machine learning techniques, including back propagation neural networks and probabilistic neural networks, to identify pests and diseases in coconut plants. Morphological feature extraction was applied in the study with accuracy around 100% in pest and disease identification. It highlights the rich possibility of combining multispectral imaging with advanced machine learning techniques to guide disease management in coconut cultivation. In another study, Rodríguez et al. ( 2021 ) also examined the damage to coconut plants due to Typhoon Goni using Sentinel-2 multispectral imagery. They conducted analysis about vegetation changes with a 90% accuracy in identifying areas affected. The high accuracy demonstrates that multispectral imaging may be used to reveal the extent of coconut crop stress by showing the remarkable details of coconut crop health. In addition, the multispectral object detection study by Gani et al. ( 2021 ) demonstrates that deep learning models can significantly improve disease detection precision in agricultural settings. Although this study focused less specifically on disease detection of coconut, it observed how the accuracy increased by adding multispectral data for object detection tasks. Based on these results, similar methodologies might be adapted for the detection of coconut disease with high accuracy. Divyanth et al. ( 2022 ) also create an attention guided Faster R-CNN model to identify coconut clusters in occlusion situations. This approach reaches a mean average precision (mAP) of 0.88 with individual class accuracies of 0.91, 0.90, 0.86 and 0.85 respectively. Thus, this study shows the power of deep learning frameworks on allowing the harvesting process to be more accurate in occluded scenes of agricultural areas. The main problem is detection of Weligama Wilt Disease (WCLWD) and the sensitivity and scalability of the current research method. Data from the early stages of the diseases are used in recently published works based on the method of spectral analysis with relatively low sensitivity. For instance, calibration of plant diseases by spectral analysis has been applied to diagnose the disease but is less sensitive to the early stages of disease, as discussed in the studies of Ahmadi et al. ( 2022 ) and Liaghat et al. ( 2014 ). Further, many of these approaches are not scalable and thus not likely viable for mass coconut lands where action needs to be prompt, for instance for disease outbreaks (Ahmadi et al. 2022 ). Additionally, there are no studies on the optimal spectral band combination for WCLWD identification. To manage these critical gaps, the current study seeks to integrate the modern advanced unmanned aerial vehicle multispectral imaging and Object-Based Image Analysis (OBIA). Through combining these technologies, the proposed method improves the capability for detecting WCLWD in large plantations both in terms of sensitivity and scalability. Multispectral imagery using UAV provides high-resolution and large-extent data acquisition of the field, and the first signs of diseases that are hardly discernible by the human eye can be distinguished (Ahmadi et al. 2022 ). Furthermore, the OBIA approach allows identification of objects within the image and the spatial pattern and relationship, which can greatly enhance the classification accuracy over more traditional pixel-based methods (Patrício and Rieder 2018 ). Materials and methods Materials Imaging system Data acquired from UAV with a multispectral camera with 6 bands and D-RTK2 mobile station (Fig. 3 ) were used in this study. The images were taken on August 24, 2023, during 11:30 am to 12:00 am. Data was acquired using the DJI P4 Multispectral drone, which has a stabilized camera model mounted on it and a D-RTK2 mobile station. The camera produces images of 6 lenses (RGB, blue, green, red, red edge and NIR) that are specifically suitable for studying vegetation. The image resolution (pixel size) at the typical flying height of 50m is 2cm/pixel. In this study, a single flight at a 50 m flying height above the ground had a coverage area of 1 acre and produceed 70 images under standard operating conditions. The app controls the UAV and the camera during the flight and records the GPS coordinates and timestamps of each image. Study area The study was carried out in Kotavila South (5.961780, 80.480186), Kamburugamuwa, Galle District, Southern Province of Sri Lanka. Where WCLWD is prevalent, and coconut trees are abundant. now developed into a quarantine disease in Sri Lanka, particularly affecting coconut production higher in the Southern Province alone, where over 40,000 ha are affected (De Silva et al. 2023 ). The fact that the diseases in question thrive within the climatic conditions and agricultural practices of this region shows that the region is definitely an area of focus in terms of disease research and management (De Silva et al. 2021 ). The high coconut plant density in Kotavila South shows that the disease environment is ideal to sustain WCLWD and proves the need for proper monitoring and performing various techniques like UAV-based multispectral imaging to manage the disease effectively (Kumara et al. 2015 ). Figure 4 depicts the location of Kotavila South, Kamburugamuwa, in Sri Lanka. While depicted, the coconut cultivation land was subjected to this study. Methods The methodology of the study consists of image acquisition, image preprocessing, image segmentation, image classification and accuracy assessment. Figure 5 illustrates the overview of methodology. Preprocessing To obtain accurate spatial information about the UAV images, georeferencing and orthorectification is done using Pix4Dmapper. This process effectively removes the distortions that are introduced by the angle and slope of shooting and gives a true metric representation that can be effectively used for previewing the form of the coconut plantations (Guo et al. 2019 ). After this, radiometric calibration is done in order to adjust the image brightness and its reflectance values towards a standard for various lighting (Tu et al. 2019 ). Band combination The combination of five spectral bands—blue (B), green (G), red (R), red edge (RE), and near-infrared (NIR)—is performed using ArcMap. Table 1 demonstrates the various band combinations. Table 1 individual band and different band combinations. Individual-band Two-band combinations Three-band combinations Four-band combinations Five-band combinations B G R RE NIR B + G B + R B + RE B + NIR G + R G + RE G + NIR R + RE R + NIR RE + NIR B + G + R B + G + RE B + G + NIR B + R + RE B + R + NIR B + RE + NIR G + R + RE G + R + NIR R + RE + NIR G + RE + NIR B + G + R + RE B + G + R + NIR B + G + RE + NIR B + R + RE + NIR G + R + RE + NIR B + G + R + RE + NIR Image segmentation Image segmentation is a critical methodology in identifying WCLWD using UAV multispectral images. This technique increases the accuracy of disease identification as it involves performing region of interest based on spectral signatures, which represent plant health and stress (Martínez-Casasnovas et al. 2021 ; Mia et al. 2023 ). Object-based image classification enables the contextual information, improving classification accuracy compared to that of the pixel-based approach (Ruwaimana et al. 2018 ; De Castro et al. 2021 ). Image Classification . This approach allows separation of important spectral features that need to discriminate between healthy and infected coconut palms by WCLWD. Reflection spectra of the leaves are recorded and disease symptoms on the plant associated with phytoplasma infection, such as chlorosis and necrosis are identified based on specific wavelengths (De Silva et al. 2021 , 2023 ). The model is trained on classified data and is able to make correlations with healthy and diseased palms. Model Evaluation As a performance indicator, the confusion matrix generated using the ENVI software is used in the method for the assessment of a classification accuracy of the model for detection of WCLWD, based on the object based image classification of UAV multispectral images. The classification of the test results can be observed by a confusion matrix containing actual positive, actual negative, predicted positive, and predicted negative and these are useful to calculate several performance measures: They include accuracy, precision, recall/sensitivity, and specificity (Hejmanowska et al. 2021 ; Riehl et al. 2023 ). This will mean that the accuracy of the classified images can be compared with ground truth to obtain a measure of the classification accuracy (Heydarian et al. 2022 ). Moreover, the way of presenting the confusion matrix might also be improved, so that it will better be interpreted and distinguished between misclassification and the assessment of the global performance of the model (García-Balboa et al. 2018 ; Luque et al. 2022 ). Table 2 provides an interpretation of the Kappa coefficient values for the classification accuracy classes (Viera and Garrett 2005 ). Table 2 Kappa coefficient description Kappa coefficient Description < 0 Less than chance agreement 0.01–0.20 Slight agreement 0.21–0.40 Fair agreement 0.41–0.60 Moderate agreement 0.61–0.80 Substantial agreement 0.81–0.99 Almost perfect agreement Results Image acquisition The image acquisition was performed in two flights, covering about 1.17 ha of the study area. The total flight time was about 30 minutes, and the total number of images was 1332. Figure 6 shows the flight planning, while Fig. 7 shows samples of acquired spectral images including red edge, NIR, green, RGB, red, and blue. Pre-processing Output The images obtained from the UAV were first corrected for their radiometric values and were georeferenced with the Rectified Skewed Orthomorphic (RSO) projection. After georeferencing, orthorectification was done to correct for skewness in the images so as to correctly represent the spatial aspect. This allows for the right alignment of every single band with geographical coordinates, making for more or less accurate analysis possible. Associated Orthorectified images were mosaiced to produce a general representation of the study area to perform the overall analysis for the multispectral data. Ground Census Data on occurrence and severity of WCLWD in coconut palms were collected by ground census in the study area. The census was carried out in association with the Coconut Research Institute (CRI) of Sri Lanka, which supplied the field equipment and technical expertise. In the study area, the census involved individual visits to every coconut palm for visual inspection and laborory testing for symptoms of WCLWD, such as yellowing, wilting, and necrosis leaves, stem bleeding and rotting. Yellow labels were applied to the diseased palms and their coordinates noted using a handheld GPS unit. The results from the multispectral imagery analysis were validated using ground census data; predictive models for WCLWD using object based image analysis (OBIA) were developed. Figure 8 is the ground census data plotted with ArcMap software and digitised. Segmentation These pre-processed images were processed by the OBIA applying 5 single bands and 26 band combinations. Each image combination was processed with a watershed segmentation using ENVI 5.0 software. To find out the best parameters related to the Segment and Merging processes, the trial and error method were conducted. The findings showed that the optimal Segment level was Edge-based at a Scale Level 50 while the Merging remained optimal at a level of 20 with the use of the FLS algorithm, this with a TKS of 3. These values were selected because they allowed for distinguishing the coconut tree canopy from fronds and minimized over segmentation problems. The segmented images were then utilized for further classification employing SVM using the four categorized segmented images. SVM Classification For the SVM classification, 38 and 7 coconut palms for normal and diseased trees were selected as training samples. The training samples were randomly collected and well covered across the study area. SVM was used instead of other classifiers due to various studies calling for a better classification accuracy, particularly for multispectral images (Pal and Mather 2005; Ballanti et al. 2016). The SVM classification results were illustrated in pseudo-color images with green color representing unaffected trees and red color for diseased trees. The SVM classifiers were optimized using a trial-and-error method whereby the optimum classification accuracy was achieved. The parameters utilized in the current research were kernel type as Radial Basis Function (RBF) kernel type and gamma in kernel function as 0.333 with 100 parameters being imposed to Germ. The Degree of Kernel Polynomial (DoKP) was adjusted at 1 while the Bias in Kernel Function (BiKF) was also adjusted at 1. The converging value was defined as 5, which equals a 95% confidence level for classification purposes. The analysis also put the unclassified in another class so as to reduce the level of error in the eventual classification. The accuracy of the classification outputs was evaluated by comparing the OBIA output with the ground census and calculating the percentage of accuracy and kappa coefficient from the confusion matrix. Table 3 illustrates the accuracy rankings and the corresponding Kappa coefficients for each combination. Table 3 Accuracy Assessment of Classification Output for Individual Band and Different Band Combinations. Band combinations Rank Overall accuracy (% ) Kappa coefficient B 28 65.01 0.3453 G 26 67.19 0.3466 R 27 67.13 0.377 RE 22 71.89 0.407 NIR 31 53.88 0.1194 B + G 21 72.5 0.42 B + R 20 74.43 0.43 B + RE 13 76.11 0.45 B + NIR 25 68.07 0.3 G + R 30 55.12 0.1282 G + RE 19 74.69 0.44 G + NIR 17 75.34 0.4512 R + RE 29 59.14 0.225 R + NIR 10 76.75 0.4725 RE + NIR 24 70.31 0.35 B + G + R 7 78.12 0.48 B + G + RE 3 79.04 0.49 B + G + NIR 9 77.21 0.475 B + R + RE 11 76.5 0.46 B + R + NIR 13 76.22 0.455 B + RE + NIR 16 75.51 0.44 G + R + RE 18 75.15 0.4788 G + R + NIR 6 78.12 0.4838 R + RE + NIR 23 71.67 0.4109 G + RE + NIR 2 79.12 0.4706 B + G + R + RE 5 78.5 0.485 B + G + R + NIR 4 78.75 0.48 B + G + RE + NIR 1 79.25 0.493 B + R + RE + NIR 11 76.5 0.46 G + R + RE + NIR 15 75.75 0.445 B + G + R + RE + NIR 8 77.31 0.485 Discussion This study effectively illustrates the impact of various band combinations on classification accuracy. The integration of diverse spectral bands leads to significant accuracy enhancement, evidenced by the combination of B + G + RE + NIR, which has the highest overall accuracy of 79.25% and a Kappa coefficient of 0.493, than other combinations. Comparatively, the simplest band combinations, such as individual bands or pairs like B and NIR, yielded lower accuracy, with B alone at 65.03% and NIR at 53.88%. When comparing three-band combinations B + G + R, an accuracy of 78.0% was achieved with a Kappa coefficient of 0.48. B + G + RE, Slightly higher accuracy at 79.0%, Kappa coefficient of 0.49. G + R + NIR, achieved a 78.12% accuracy with a Kappa coefficient of 0.4838. Four-band combinations generally performed better than simpler combinations. B + G + R + RE, Achieved an accuracy of 78.5% and a Kappa coefficient of 0.485. B + G + R + NIR, Higher accuracy at 78.75% with a Kappa coefficient of 0.48. Figure 10 illustrates the classification results, identifying both normal trees and those infected with WCLWD by utilizing a combination of Blue, Green, Red Edge, and Near-Infrared (B + G + RE + NIR) spectral bands. In Fig. 11 , a comparative visualization of the classification results is provided, with normal trees denoted by black circles and infected trees depicted with white circles. Previous works have also reported that the SVM classifier proves to be better than other classifiers like KNN and RF for plant disease detection. For example, Chen found that when developing plant disease classifiers, the SVM classifier had precision higher than 97% while the RF was lower (Chen et al. 2023 ). The study of Rahayu F. et al. (2021) also indicates that for texture and color features, SVM yields better results compared to KNN and DT to classify the diseased leaves. These together highlight that SVM is much more accurate in plant disease detection tasks than the KNN and RF classifiers. Also Support Vector Machine (SVM) classification effectively improves the ability to classify vegetation into different classes by using its spectral traits. SVM has been employed in numerous research to distinguish the various types of plants; it has also been utilized in determining health standards Since; it is capable of analyzing numerous fields of understanding, it is effective regardless the size of the data set Essay on SVM (Buters et al. 2019 ). The study highlights that NIR band important in remote sensing applications since the combinations with NIR produced better results throughout the assessments and demonstrated the efficiency of NIR in improving the outcomes of classification. The classification results in the near-infrared (NIR) region have been shown to increase by offering useful information about plant health and stress conditions (Hall et al. 2018 ). The NIR band is particularly important as it is sensitive to variations in leaf chlorophyll content, which can indicate disease presence (Hu et al. 2020 ). Analysis of the data showed that different spectral band imposed different flexibility on classification of images to changes in the biophysical characteristics of plants, in particular nitrogen and chlorophyll content. However, as studies have suggested even stronger sensitivity of the G band at canopy level than that of the R band, which rises with chlorophyll content (Fernández et al. 2019 ). Instead, the R band is targeted more at chlorophyll a and b, which are very important in the photosynthesis process and can be quantified using the spectral indices (Lohner et al. 2022 ). Previous studies have demonstrated that the potential of the RE band for physiological conditions of vegetation are consistent with studies reporting that this band is more sensitive to stress and chlorophyll content of plant canopies (Hunt et al. 2011 ; Adelabu et al. 2014 ). Moreover, the near-infrared (NIR) band is very efficient for obtaining data from moisture status and moisture content in the plant, other plant stress and biomass characterization features (Verrelst et al. 2016 ). Consequently, we noticed the highest accuracy of each entire band in the B band 65.01%, RE band 71.89%, G band 67.19%, R band 67.13%, and NIR band 53.88%. Our findings also revealed that stress on plants and the reduction in chlorophyll a and b, as indicated by the RE and R bands were critical factors that influenced disease condition, seconded by plant/leaf nitrogen and pigment loss as indicated by the G band. Based on an integration of NIR and RE bands, great capacity with the disease diagnosis indications processing has been established. The concurrence of these bands has been found to increase differentiation of disease severity of vegetative signs, increasing diagnostic effectiveness (Heim et al. 2019 ; Liu et al. 2021 ). In particular, the application of multispectral imaging has been discovered to provide positive outcomes in disease identification in agricultural crops since spectral indices were used successfully to detect disease symptoms in various crops (Mahlein et al. 2010 ). In addition, UAV-based multispectral images are also useful for disease diagnosis in agriculture and provide a basis for effective plant health diagnosis and disease diagnosis in agriculture, which provides a good basis for a combination of NIR and RE bands for better identification (Liang et al. 2019 ; Tait et al. 2019 ). Nainanayake et al. ( 2016a ) created a modified vegetation index including the R, G, and twice B bands and developed an indicative technique that could improve the identification of coconut tree disease with an accuracy of more than 80%. Meanwhile, (Xavier et al. 2019 ) identified four stages of Ramulia leaf blight disease of cotton using G, R, and NIR bands of the image separately. It can differentiate early, severe conditions, stated they, but it can’t be used to differentiate healthy from severe, and healthy from mild. The difficulty in discriminating trees from the ground is the most challenging undertaking experienced in this study, especially in very dense vegetated areas. Because of the canopy structure of complex canopies and of foliage overlapping, it is hard to accurately classify trees individually. These issues can be overcome using object based image analysis (OBIA) techniques by segmenting the images into meaningfully segmented objects rather than performing pixel based classifications. The use of this approach helps enhances classification outcomes by providing an understanding of the spatial relationships between trees and their surroundings (Buters et al. 2019 ). Furthermore, the presence of other diseases, environmental stresses, and pest infestations, as well as all the other factors known to affect crop productivity tends to affect the reflectance values selected from the multispectral images, hence compromising the accuracy of the classifications. Reflectance values depend on a number of biotic and abiotic factors that can cause similarity between the symptoms of different diseases. For instance, pest damage leaves’ reflectance and environmental stresses such as drought and nutrient availability can distort reflectance characteristics, which makes it hard to differentiate between healthy or diseased plants only from the spectral data (Abdulridha et al. 2020 ; Zhang et al. 2022 ) . Further research should be done to overcome the difficulties in distinguishing symptom overlap from those of other diseases or pests. This could involve the study of hyperspectral imaging, which offers a wider spectrum, that is, it probably covers a wider spectral region resolution, which could help distinguish plant health status. Additionally, the utilization of multispectral data in conjunction with other more higher level of machine learning algorithms can lead to an even better identification of disease present in coconut trees (Loladze et al. 2019 ). Conclusion In this study detect WCLWD-infected palms, the multispectral images of the plantation obtained with the use of the UAV were subjected to OBIA analysis, including segmentation and merging with the SVM classifier. Thus, the results indicated that different band combinations and individual band accuracy varied, highlighting the importance of selecting the optimal combination for accurate analysis. It was moderate and ranged from 53.88 to 71.89 regarding the accuracy of the individuals’ band’s performances. Conversely, band image combinations of more than three bands. Further confirming the findings of other researchers, with the maximum accuracy 79.25%, arising from B _ G _ RE _NIR. This shows that OBIA analysis with a four-band combination of multispectral images can be used to detect coconut trees that are infected with WCLWD. Accurate classification and interpretation of the data is still a challenge due to variation in environmental factors and confusing signs similar to other coconut diseases. Therefore, this study recommends that a combination of molecular biology, genetics, and other relevant imaging techniques is useful and needed to further design the management techniques of WCLWD. Further study employing an adequate OBIA analysis experiment that incorporates sophisticated classifiers like support vector machine (SVM) and random forest (RF) alongside a complex template match algorithm while trying to detect early WCLWD infection in coconut palms using multispectral band combinations and hyperspectral images should be done. Explore temporal aspects of WCLWD. In what way does there occur alteration in spectral signatures due to disease state change? It is recommended that future research incorporate advanced methodologies, such as molecular biology and genetics, alongside imaging techniques to develop comprehensive management strategies for WCLWD. In terms of practical recommendations for plantation managers, it is crucial to emphasize the integration of UAV imaging with ground surveys to facilitate early detection of WCLWD. This combined approach can enhance the reliability of disease monitoring and improve management practices. Additionally, the methodologies developed for WCLWD detection could be adapted for monitoring other agricultural diseases or environmental stress factors, thereby broadening the impact of this research in precision agriculture. Declarations Compliance with ethical standards Conflict of interest The author declares that they have no confict of interest. References Abdulridha J, Ampatzidis Y, Qureshi J, Roberts P (2020) Laboratory and UAV-Based Identification and Classification of Tomato Yellow Leaf Curl, Bacterial Spot, and Target Spot Diseases in Tomato Utilizing Hyperspectral Imaging and Machine Learning. Remote Sens 12:2732. https://doi.org/10.3390/rs12172732 Adelabu S, Mutanga O, Adam E (2014) Evaluating the impact of red-edge band from Rapideye image for classifying insect defoliation levels. ISPRS J Photogramm Remote Sens 95:34–41. https://doi.org/10.1016/j.isprsjprs.2014.05.013 Ahmadi P, Mansor S, Farjad B, Ghaderpour E (2022) Unmanned Aerial Vehicle (UAV)-Based Remote Sensing for Early-Stage Detection of Ganoderma. Remote Sens 14:1239. https://doi.org/10.3390/rs14051239 Ballanti L, Blesius L, Hines E, Kruse B (2016) Tree Species Classification Using Hyperspectral Imagery: A Comparison of Two Classifiers. Remote Sens 8:445. https://doi.org/10.3390/rs8060445 Barman U, Pathak C, Mazumder NK (2023) Comparative assessment of Pest damage identification of coconut plant using damage texture and color analysis. Multimed Tools Appl 82:25083–25105. https://doi.org/10.1007/s11042-023-14369-2 Buters T, Belton D, Cross A (2019) Seed and Seedling Detection Using Unmanned Aerial Vehicles and Automated Image Classification in the Monitoring of Ecological Recovery. Drones 3:53. https://doi.org/10.3390/drones3030053 Chen H, Han Y, Liu Y et al (2023) Classification models for Tobacco Mosaic Virus and Potato Virus Y using hyperspectral and machine learning techniques. Front Plant Sci 14:1211617. https://doi.org/10.3389/fpls.2023.1211617 De Castro AI, Shi Y, Maja JM, Peña JM (2021) UAVs for Vegetation Monitoring: Overview and Recent Scientific Contributions. Remote Sens 13:2139. https://doi.org/10.3390/rs13112139 De Silva PHPR, Aratchige NS, Ranasinghe CS, Kantha AAFL (2021) Diseased palm removals as a strategy for the successful managementof Weligama coconut leaf wilt phytoplasma disease of coconut in SriLanka. Phytopathogenic Mollicutes 11:79–85. https://doi.org/10.5958/2249-4677.2021.00013.X De Silva PR, Perera CN, Bahder BW, Attanayake RN (2023) Nested PCR-Based Rapid Detection of Phytoplasma Leaf Wilt Disease of Coconut in Sri Lanka and Systemic Movement of the Pathogen. Pathogens 12:294. https://doi.org/10.3390/pathogens12020294 Divyanth LG, Soni P, Pareek CM et al (2022) Detection of Coconut Clusters Based on Occlusion Condition Using Attention-Guided Faster R-CNN for Robotic Harvesting. Foods 11:3903. https://doi.org/10.3390/foods11233903 Fernández E, Gorchs G, Serrano L (2019) Use of consumer-grade cameras to assess wheat N status and grain yield. PLoS ONE 14:e0211889. https://doi.org/10.1371/journal.pone.0211889 Gani MO, Kuiry S, Das A et al (2021) Multispectral Object Detection with Deep Learning. In: Dutta P, Mandal JK, Mukhopadhyay S (eds) Computational Intelligence in Communications and Business Analytics. Springer International Publishing, Cham, pp 105–117 García-Balboa JL, Alba-Fernández MV, Ariza-López FJ, Rodríguez-Avi J (2018) Analysis of Thematic Similarity Using Confusion Matrices. ISPRS Int J Geo-Inf 7:233. https://doi.org/10.3390/ijgi7060233 Guo Y, Senthilnath J, Wu W et al (2019) Radiometric Calibration for Multispectral Camera of Different Imaging Conditions Mounted on a UAV Platform. Sustainability 11:978. https://doi.org/10.3390/su11040978 Hall O, Dahlin S, Marstorp H et al (2018) Classification of Maize in Complex Smallholder Farming Systems Using UAV Imagery. Drones 2:22. https://doi.org/10.3390/drones2030022 Heim R, Wright I, Scarth P et al (2019) Multispectral, Aerial Disease Detection for Myrtle Rust (Austropuccinia psidii) on a Lemon Myrtle Plantation. Drones 3:25. https://doi.org/10.3390/drones3010025 Hejmanowska B, Kramarczyk P, Głowienka E, Mikrut S (2021) Reliable Crops Classification Using Limited Number of Sentinel-2 and Sentinel-1 Images. Remote Sens 13:3176. https://doi.org/10.3390/rs13163176 Heydarian M, Doyle TE, Samavi R (2022) MLCM: Multi-Label Confusion Matrix. IEEE Access 10:19083–19095. https://doi.org/10.1109/ACCESS.2022.3151048 Hu X, Yang L, Zhang Z (2020) Non-destructive identification of single hard seed via multispectral imaging analysis in six legume species. Plant Methods 16:116. https://doi.org/10.1186/s13007-020-00659-5 Hunt ER, Daughtry CST, Eitel JUH, Long DS (2011) Remote Sensing Leaf Chlorophyll Content Using a Visible Band Index. Agron J 103:1090–1099. https://doi.org/10.2134/agronj2010.0395 Kanatiwela-de Silva C, Damayanthi M, De Silva N et al (2019) Immunological detection of the Weligama coconut leaf wilt disease associated phytoplasma: Development and validation of a polyclonal antibody based indirect ELISA. PLoS ONE 14:e0214983. https://doi.org/10.1371/journal.pone.0214983 Kumara ADNT, Chandrashekharaiah M, Kandakoor SB, Chakravarthy AK (2015) Status and Management of Three Major Insect Pests of Coconut in the Tropics and Subtropics. In: Chakravarthy AK (ed) New Horizons in Insect Science: Towards Sustainable Pest Management. Springer India, New Delhi, pp 359–381 Liaghat S, Ehsani R, Mansor S et al (2014) Early detection of basal stem rot disease ( Ganoderma ) in oil palms based on hyperspectral reflectance data using pattern recognition algorithms. Int J Remote Sens 35:3427–3439. https://doi.org/10.1080/01431161.2014.903353 Liang W, Zhang H, Zhang G, Cao H (2019) Rice Blast Disease Recognition Using a Deep Convolutional Neural Network. Sci Rep 9:2869. https://doi.org/10.1038/s41598-019-38966-0 Liu J, Xiang J, Jin Y et al (2021) Boost Precision Agriculture with Unmanned Aerial Vehicle Remote Sensing and Edge Intelligence: A Survey. Remote Sens 13:4387. https://doi.org/10.3390/rs13214387 Lohner SA, Biegert K, Hohmann A et al (2022) Chlorophyll- and anthocyanin-rich cell organelles affect light scattering in apple skin. Photochem Photobiol Sci 21:261–273. https://doi.org/10.1007/s43630-021-00164-1 Loladze A, Rodrigues FA, Toledo F et al (2019) Application of Remote Sensing for Phenotyping Tar Spot Complex Resistance in Maize. Front Plant Sci 10:552. https://doi.org/10.3389/fpls.2019.00552 Luque A, Mazzoleni M, Carrasco A, Ferramosca A (2022) Visualizing Classification Results: Confusion Star and Confusion Gear. IEEE Access 10:1659–1677. https://doi.org/10.1109/ACCESS.2021.3137630 Mahlein A-K, Steiner U, Dehne H-W, Oerke E-C (2010) Spectral signatures of sugar beet leaves for the detection and differentiation of diseases. Precis Agric 11:413–431. https://doi.org/10.1007/s11119-010-9180-7 Martínez-Casasnovas JA, Sandonís-Pozo L, Escolà A et al (2021) Delineation of Management Zones in Hedgerow Almond Orchards Based on Vegetation Indices from UAV Images Validated by LiDAR-Derived Canopy Parameters. Agronomy 12:102. https://doi.org/10.3390/agronomy12010102 Mia MS, Tanabe R, Habibi LN et al (2023) Multimodal Deep Learning for Rice Yield Prediction Using UAV-Based Multispectral Imagery and Weather Data. Remote Sens 15:2511. https://doi.org/10.3390/rs15102511 Nainanayake AD, Gunathilake J, Kumarathunga MDP et al (2016a) Limitation in the use of spectral analysis to detect Weligama coconut leaf wilt disease affected palms in Southern Sri Lanka. COCOS 22:13–24. https://doi.org/10.4038/cocos.v22i1.5808 Nainanayake AD, Kumarathunga MDP, De Silva PHPR (2016b) A survey of land for Weligama coconut leaf wilt disease affected palms outside the declared boundary in the Southern Province. COCOS 22:57–64. https://doi.org/10.4038/cocos.v22i1.5812 Pal M, Mather PM (2005) Support vector machines for classification in remote sensing. Int J Remote Sens 26:1007–1011. https://doi.org/10.1080/01431160512331314083 Patrício DI, Rieder R (2018) Computer vision and artificial intelligence in precision agriculture for grain crops: A systematic review. Comput Electron Agric 153:69–81. https://doi.org/10.1016/j.compag.2018.08.001 Perera SACN, Herath HMNB, Wijesekera HTR et al (2016a) Evaluation of coconut germplasm in Weligama and Matara area of the Southern Province of Sri Lanka for resistance to Weligama coconut leaf wilt disease. COCOS 21:15–20. https://doi.org/10.4038/cocos.v21i0.5803 Perera SACN, Kumarasinghe WM, Gunasekara TMCP (2016b) Assessing the performance of fruit colour based phenotypes of tall (Typica) coconuts (Cocos nucifera L.) in Sri Lanka. COCOS 22:25–29. https://doi.org/10.4038/cocos.v22i1.5809 Prades A, Salum UN, Pioch D (2016) New era for the coconut sector. What prospects for research? OCL 23:D607. https://doi.org/10.1051/ocl/2016048 Rahayu FBR, Mudjirahardjo P, Muslim MA (2021) Leaf Diseases Classification on Peanut Leaves Based on Texture and Colour Features. Int J Comput Appl Technol Res 10:149–155. https://doi.org/10.7753/IJCATR1006.1004 Rajan P (2011) Gradient Outbreak of Coconut Slug Caterpillar, Macroplectra nararia Moore in East Coast of India. CORD 27:7. https://doi.org/10.37833/cord.v27i1.124 Riehl K, Neunteufel M, Hemberg M (2023) Hierarchical confusion matrix for classification performance evaluation. J R Stat Soc Ser C Appl Stat 72:1394–1412. https://doi.org/10.1093/jrsssc/qlad057 Rodríguez AC, Daudt RC, D’Aronco S et al (2021) Robust Damage Estimation of Typhoon Goni on Coconut Crops with Sentinel-2 Imagery. Remote Sens 13:4302. https://doi.org/10.3390/rs13214302 Ruwaimana M, Satyanarayana B, Otero V et al (2018) The advantages of using drones over space-borne imagery in the mapping of mangrove forests. PLoS ONE 13:e0200288. https://doi.org/10.1371/journal.pone.0200288 Subajiny S, Dilini B, Terrence M (2018) A comparative study on stability of different types of coconut (Cocos nucifera) oil against autoxidation and photo-oxidation. Afr J Food Sci 12:216–229. https://doi.org/10.5897/AJFS2018.1695 Tait L, Bind J, Charan-Dixon H et al (2019) Unmanned Aerial Vehicles (UAVs) for Monitoring Macroalgal Biodiversity: Comparison of RGB and Multispectral Imaging Sensors for Biodiversity Assessments. Remote Sens 11:2332. https://doi.org/10.3390/rs11192332 Tu Y-H, Johansen K, Phinn S, Robson A (2019) Measuring Canopy Structure and Condition Using Multi-Spectral UAS Imagery in a Horticultural Environment. Remote Sens 11:269. https://doi.org/10.3390/rs11030269 Verrelst J, Rivera JP, Gitelson A et al (2016) Spectral band selection for vegetation properties retrieval using Gaussian processes regression. Int J Appl Earth Obs Geoinf 52:554–567. https://doi.org/10.1016/j.jag.2016.07.016 Viera A, Garrett J (2005) Understanding Interobserver Agreement: The Kappa Statistic. Fam Med 37:360–363 Wijesekara HTR, Perera SACN, Bandupriya D et al (2020) Detection of Weligama Coconut Leaf Wilt Disease Phytoplasma by Real-Time Polymerase Chain Reaction. CORD 36:1–5. https://doi.org/10.37833/cord.v36i.425 Wijesekara T, Perera L, I. W, et al (2008) Preliminary Investigation on Weligama Coconut leaf Wilt disease: A new disease in southern part of Sri Lanka Xavier TWF, Souto RNV, Statella T et al (2019) Identification of Ramularia Leaf Blight Cotton Disease Infection Levels by Multispectral, Multiscale UAV Imagery. Drones 3:33. https://doi.org/10.3390/drones3020033 Zhang S, Li X, Ba Y et al (2022) Banana Fusarium Wilt Disease Detection by Supervised and Unsupervised Methods from UAV-Based Multispectral Imagery. Remote Sens 14:1231. https://doi.org/10.3390/rs14051231 Cite Share Download PDF Status: Published Journal Publication published 12 Jul, 2025 Read the published version in Journal of Plant Diseases and Protection → Version 1 posted Reviewers agreed at journal 31 Jan, 2025 Reviewers invited by journal 29 Jan, 2025 Editor invited by journal 21 Jan, 2025 Editor assigned by journal 20 Jan, 2025 First submitted to journal 09 Jan, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5767642","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":408556868,"identity":"1c946b0f-ce6b-478c-91d1-2ffd850408bf","order_by":0,"name":"H.D.M.U Wijesinghe1 H.D.M.U Wijesinghe1","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"H.D.M.U","middleName":"Wijesinghe1 H.D.M.U","lastName":"Wijesinghe1","suffix":""},{"id":408556869,"identity":"516fcfaf-0315-4c13-8c2a-2bdd2c5a29fc","order_by":1,"name":"KMC Tahrupath","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"KMC","middleName":"","lastName":"Tahrupath","suffix":""},{"id":408556870,"identity":"5c0a4be2-6537-4ffd-aac9-9041887c9c4c","order_by":2,"name":"JAYASINGHE GUTTILA","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA90lEQVRIiWNgGAWjYDACZjBpwcPAwH7gQIKBDZDD2HiAgBbGBgYGCaAWnsQHHyrSQFoa8GthgGgBMYwNZ5w5DBbCq8XgOPPzBx/3SMjwSzekSfO2nbdb234YaEuNTTROLYfZDBtnPJPgkZxz8BhQy+3kbWcSgVqOpeU24NAi2czD2MxzQILH4EZCGliL2QGgFsaGw0RpMQNqOZdsdv4hfi38zAgtIO8fsDO7QcAWfmY2w5kzgFokZ+SAAjk5wewG0JYEPH5h4z/84MOHAzb2/BLpoKi0szc7n/7wwYcaG5xaMEAiWGUCscpBwJ4UxaNgFIyCUTAyAAC++GEau4wWzAAAAABJRU5ErkJggg==","orcid":"","institution":"","correspondingAuthor":true,"prefix":"","firstName":"JAYASINGHE","middleName":"","lastName":"GUTTILA","suffix":""}],"badges":[],"createdAt":"2025-01-05 12:05:00","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5767642/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5767642/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s41348-025-01115-z","type":"published","date":"2025-07-12T15:57:03+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":75407880,"identity":"28a7a016-cbd3-4bbd-bd3b-26d5501c9aa0","added_by":"auto","created_at":"2025-02-04 08:55:50","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":663515,"visible":true,"origin":"","legend":"\u003cp\u003eProgressive development of symptoms in coconut palms affected by WCLWD 1. Flaccidity, 2. Yellowing, 3. Drying of leaflets from margin, 4. Breaking tips of fronds\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/15558d9c0b426855a9ca777d.jpg"},{"id":75407878,"identity":"1ed486cd-97d7-4062-ba93-25a801dec8a1","added_by":"auto","created_at":"2025-02-04 08:55:50","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":117988,"visible":true,"origin":"","legend":"\u003cp\u003eSymptoms of root system of palms affected by WCLWD a) Necrosis in root area, b) Extensive branching\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/8f94d3c2c283540c4e25a9dd.jpg"},{"id":75409454,"identity":"f053b9b3-12e3-4cee-a7b0-2629ab6b35fb","added_by":"auto","created_at":"2025-02-04 09:03:50","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":468797,"visible":true,"origin":"","legend":"\u003cp\u003eImaging setup; a) DJI P4 drone mounted with a Multispectral camera and D-RTK2 mobile station. b) DJI P4 Multispectral drone with multispectral camera\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/130e1ceacb8c589a54bf75b8.jpg"},{"id":75407885,"identity":"7d1981cf-99dc-4420-936f-e09a1fb0b0d2","added_by":"auto","created_at":"2025-02-04 08:55:50","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":409513,"visible":true,"origin":"","legend":"\u003cp\u003eThe location map of the study area, Waligama, Matara, Sri Lanka\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/a0cacb6fde3eadcfc4191e2b.jpg"},{"id":75407901,"identity":"7c67a16a-7172-4973-8dfe-29089cb9ea17","added_by":"auto","created_at":"2025-02-04 08:55:51","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":66605,"visible":true,"origin":"","legend":"\u003cp\u003eWorkflow for Detecting Weligama Wilt Disease in Coconut Using UAV Multispectral Imaging and Object-Based Classification\u003c/p\u003e","description":"","filename":"Picture5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/f0f2754d536ee7b7896c1c53.jpg"},{"id":75407881,"identity":"6062f346-4c84-4ab4-bd9c-6680dc0ee3a5","added_by":"auto","created_at":"2025-02-04 08:55:50","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":111840,"visible":true,"origin":"","legend":"\u003cp\u003eFlight planning\u003c/p\u003e","description":"","filename":"Picture6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/8588b6409e2d725b50d15de8.jpg"},{"id":75407886,"identity":"7c788192-f28e-4eeb-a4cc-81761fd81638","added_by":"auto","created_at":"2025-02-04 08:55:50","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":419932,"visible":true,"origin":"","legend":"\u003cp\u003eSample of acquired spectral images: 1. Red edge, 2. NIR. 3. Green 4.RGB, 5.Red, 6.Blue\u003c/p\u003e","description":"","filename":"Picture7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/3c6c053f1b53c190b0700b94.jpg"},{"id":75409453,"identity":"82d22595-d4ea-458e-9677-d4ca89ed82c2","added_by":"auto","created_at":"2025-02-04 09:03:50","extension":"jpg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":323074,"visible":true,"origin":"","legend":"\u003cp\u003ea) Ground census digitized into their respective coordinates using ArcMap software. b) Overlap map with ground census.\u003c/p\u003e","description":"","filename":"Picture8.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/8d6c6b8de2e647216fd7639a.jpg"},{"id":75409542,"identity":"106929e0-0422-467a-9e63-b744df27b109","added_by":"auto","created_at":"2025-02-04 09:03:51","extension":"jpg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":320112,"visible":true,"origin":"","legend":"\u003cp\u003eSegmentation output: a) normal image, b) segmented image\u003c/p\u003e","description":"","filename":"Picture9.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/5742eac3ea619238bfea4e52.jpg"},{"id":75407892,"identity":"c698596d-64ba-48ef-af92-d77f8909c889","added_by":"auto","created_at":"2025-02-04 08:55:50","extension":"jpg","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":253509,"visible":true,"origin":"","legend":"\u003cp\u003eClassification result of normal and WCLWD-infected trees (B + G + RE + NIR)\u003c/p\u003e","description":"","filename":"Picture10.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/98273503097dade42add9436.jpg"},{"id":75407894,"identity":"0249de00-60e5-4a72-ae53-cac76e5fc0e4","added_by":"auto","created_at":"2025-02-04 08:55:50","extension":"jpg","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":310993,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of Classification result (Black circle for normal and white circle for infected)\u003c/p\u003e","description":"","filename":"Picture11.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/2b7d62b57431e052cbbbd1dc.jpg"},{"id":86699456,"identity":"c1b19423-240b-451e-85ce-18c7c0e645e0","added_by":"auto","created_at":"2025-07-14 16:09:58","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4178313,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5767642/v1/fec0cfcb-654d-46de-a6cf-51602982c525.pdf"}],"financialInterests":"","formattedTitle":"Detecting Weligama Coconut Leaf Wilt Disease in Coconut Using UAV-Based Multispectral Imaging and Object-Based Classification","fulltext":[{"header":"Introduction","content":"\u003cp\u003eCoconut cultivation, particularly in tropical regions, is economically and culturally significant due to its diverse uses, including food, oil, fiber, and medicine. The coconut palm (\u003cem\u003eCocos nucifera\u003c/em\u003e), occasionally called the tree of life, stands out as one of the most versatile trees globally. Coconut farming is one of the most important crops in the agricultural industry all over the world, mainly in Indonesia, the Philippines and India. The coconut market around the world is becoming wider due to people\u0026rsquo;s preference for foods and products made from coconut oil, water, and more that are considered healthy products in the food chain (Prades et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Apart from being a source of income to millions of farmers, it contributes significantly to the economies of producing countries through exports and job creation (Prades et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn Sri Lankan agricultural economy the most important tree crop is coconut., covering approximately 455,000 hectares as of 2015 and producing around 3,056\u0026nbsp;million nuts annually (Subajiny et al. \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Also contributes to about 0.5% of the GDP of the country and directly and indirectly uses the 1.5\u0026nbsp;million people (Perera et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016a\u003c/span\u003e). Sri Lanka has a very good market for copra and coconut oil that are highly demanded for use locally and in export markets. The genetic stock of coconuts in the country is diverse with the Sri Lanka Tall (Typica) as the most common progeny (Perera et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016a\u003c/span\u003e). Also, the sector has some challenges that affect productivity and sustainability such as climate change, pests, and diseases among them (Rajan \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2011\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWeligama Coconut Leaf Wilt Disease (WCLWD) is a severe threat, recognized in 2006 in the region of Weligama in Southern Sri Lanka and mainly on coconut trees (Nainanayake et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2016b\u003c/span\u003e; Kanatiwela-de Silva et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The causal agent is a phytoplasma. The disease is characterized by flaccidity and marginal necrosis of leaflets, intense yellowing of the fronds, reduction of the crown size and tapering of the trunk, loss of productivity and eventual death of the palm within two years (Perera et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2016b\u003c/span\u003e). Figures\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e show symptoms of WCLWD (Wijesekara et al. \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2008\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe traditional methods of WCLWD detection are based on visual inspection, laboratory testing, and ground surveys. And among these, the most commonly used classical techniques include the nested PCR that makes it possible to quickly identify the phytoplasma causing WCLWD (Kanatiwela-de Silva et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Also demonstrated that an indirect ELISA, validated alongside PCR, achieved high accuracy (93%) and sensitivity (92.7%) for detecting WCLWD-associated phytoplasma, although specificity was lower at 79% (Kanatiwela-de Silva et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The nested PCR approach has also shown promise, with a success rate of 88% and 100% specificity in detecting phytoplasmas in coconut tissues (De Silva et al. \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Researchers have connected the phytoplasma to several environmental stresses that potentially worsen the disease\u0026rsquo;s effects (Perera et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016a\u003c/span\u003e). Uncertainty of early diagnosis of WCLWD is attributed to low titre of phytoplasma in the affected tissues (Wijesekara et al. \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Elimination of phytoplasma through organic management is not possible because there is no effective remedial measure to control phytoplasma other than developing resistant coconut cultivars (Perera et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2016a\u003c/span\u003e). Disease spread has led to quarantining, as well as the uprooting of palms as a means of preventing the further spread of the disease (Nainanayake et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2016a\u003c/span\u003e, b). Information on the molecular characteristics is very useful to determine how the associated phytoplasma is best detected and managed (Kanatiwela-de Silva et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). But conventional methods like field visits have been used for assessing the incidences of the disease across district levels; study done in 2012, showed that 65,838, 251,980, and 14,344 palms were affected in Galle, Matara, and Hambantota, respectively (Nainanayake et al. \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2016b\u003c/span\u003e). However, these methods have their drawbacks; for example, it has been realized that spectral methods such as spectral analysis have low sensitivity in detecting palm areas affected by WCLWD (Nainanayake et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2016a\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMultispectral imaging has emerged as a pivotal technology in agricultural disease detection, particularly through the utilization of unmanned aerial vehicles (UAVs). This imaging technique captures data across multiple wavelengths, enabling the assessment of plant health by analyzing spectral signatures associated with various physiological conditions.\u003c/p\u003e \u003cp\u003eBarman et al. (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) employed machine learning techniques, including back propagation neural networks and probabilistic neural networks, to identify pests and diseases in coconut plants. Morphological feature extraction was applied in the study with accuracy around 100% in pest and disease identification. It highlights the rich possibility of combining multispectral imaging with advanced machine learning techniques to guide disease management in coconut cultivation.\u003c/p\u003e \u003cp\u003eIn another study, Rodr\u0026iacute;guez et al. (\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) also examined the damage to coconut plants due to Typhoon Goni using Sentinel-2 multispectral imagery. They conducted analysis about vegetation changes with a 90% accuracy in identifying areas affected. The high accuracy demonstrates that multispectral imaging may be used to reveal the extent of coconut crop stress by showing the remarkable details of coconut crop health.\u003c/p\u003e \u003cp\u003eIn addition, the multispectral object detection study by Gani et al. (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) demonstrates that deep learning models can significantly improve disease detection precision in agricultural settings. Although this study focused less specifically on disease detection of coconut, it observed how the accuracy increased by adding multispectral data for object detection tasks. Based on these results, similar methodologies might be adapted for the detection of coconut disease with high accuracy.\u003c/p\u003e \u003cp\u003eDivyanth et al. (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) also create an attention guided Faster R-CNN model to identify coconut clusters in occlusion situations. This approach reaches a mean average precision (mAP) of 0.88 with individual class accuracies of 0.91, 0.90, 0.86 and 0.85 respectively. Thus, this study shows the power of deep learning frameworks on allowing the harvesting process to be more accurate in occluded scenes of agricultural areas.\u003c/p\u003e \u003cp\u003eThe main problem is detection of Weligama Wilt Disease (WCLWD) and the sensitivity and scalability of the current research method. Data from the early stages of the diseases are used in recently published works based on the method of spectral analysis with relatively low sensitivity. For instance, calibration of plant diseases by spectral analysis has been applied to diagnose the disease but is less sensitive to the early stages of disease, as discussed in the studies of Ahmadi et al. (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and Liaghat et al. (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Further, many of these approaches are not scalable and thus not likely viable for mass coconut lands where action needs to be prompt, for instance for disease outbreaks (Ahmadi et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAdditionally, there are no studies on the optimal spectral band combination for WCLWD identification. To manage these critical gaps, the current study seeks to integrate the modern advanced unmanned aerial vehicle multispectral imaging and Object-Based Image Analysis (OBIA). Through combining these technologies, the proposed method improves the capability for detecting WCLWD in large plantations both in terms of sensitivity and scalability. Multispectral imagery using UAV provides high-resolution and large-extent data acquisition of the field, and the first signs of diseases that are hardly discernible by the human eye can be distinguished (Ahmadi et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Furthermore, the OBIA approach allows identification of objects within the image and the spatial pattern and relationship, which can greatly enhance the classification accuracy over more traditional pixel-based methods (Patr\u0026iacute;cio and Rieder \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eMaterials\u003c/h2\u003e \u003cdiv id=\"Sec4\" class=\"Section3\"\u003e \u003ch2\u003eImaging system\u003c/h2\u003e \u003cp\u003eData acquired from UAV with a multispectral camera with 6 bands and D-RTK2 mobile station (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) were used in this study. The images were taken on August 24, 2023, during 11:30 am to 12:00 am.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eData was acquired using the DJI P4 Multispectral drone, which has a stabilized camera model mounted on it and a D-RTK2 mobile station. The camera produces images of 6 lenses (RGB, blue, green, red, red edge and NIR) that are specifically suitable for studying vegetation. The image resolution (pixel size) at the typical flying height of 50m is 2cm/pixel. In this study, a single flight at a 50 m flying height above the ground had a coverage area of 1 acre and produceed 70 images under standard operating conditions. The app controls the UAV and the camera during the flight and records the GPS coordinates and timestamps of each image.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e\n\u003ch3\u003eStudy area\u003c/h3\u003e\n\u003cp\u003eThe study was carried out in Kotavila South (5.961780, 80.480186), Kamburugamuwa, Galle District, Southern Province of Sri Lanka. Where WCLWD is prevalent, and coconut trees are abundant. now developed into a quarantine disease in Sri Lanka, particularly affecting coconut production higher in the Southern Province alone, where over 40,000 ha are affected (De Silva et al. \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The fact that the diseases in question thrive within the climatic conditions and agricultural practices of this region shows that the region is definitely an area of focus in terms of disease research and management (De Silva et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The high coconut plant density in Kotavila South shows that the disease environment is ideal to sustain WCLWD and proves the need for proper monitoring and performing various techniques like UAV-based multispectral imaging to manage the disease effectively (Kumara et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e depicts the location of Kotavila South, Kamburugamuwa, in Sri Lanka. While depicted, the coconut cultivation land was subjected to this study.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eMethods\u003c/h3\u003e\n\u003cp\u003eThe methodology of the study consists of image acquisition, image preprocessing, image segmentation, image classification and accuracy assessment. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e illustrates the overview of methodology.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003ePreprocessing\u003c/h3\u003e\n\u003cp\u003eTo obtain accurate spatial information about the UAV images, georeferencing and orthorectification is done using Pix4Dmapper. This process effectively removes the distortions that are introduced by the angle and slope of shooting and gives a true metric representation that can be effectively used for previewing the form of the coconut plantations (Guo et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). After this, radiometric calibration is done in order to adjust the image brightness and its reflectance values towards a standard for various lighting (Tu et al. \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eBand combination\u003c/h2\u003e \u003cp\u003eThe combination of five spectral bands\u0026mdash;blue (B), green (G), red (R), red edge (RE), and near-infrared (NIR)\u0026mdash;is performed using ArcMap. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e demonstrates the various band combinations.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eindividual band and different band combinations.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIndividual-band\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTwo-band combinations\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThree-band combinations\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eFour-band combinations\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eFive-band combinations\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eB\u003c/p\u003e \u003cp\u003eG\u003c/p\u003e \u003cp\u003eR\u003c/p\u003e \u003cp\u003eRE\u003c/p\u003e \u003cp\u003eNIR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;R\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eG\u0026thinsp;+\u0026thinsp;R\u003c/p\u003e \u003cp\u003eG\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e \u003cp\u003eG\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eR\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e \u003cp\u003eR\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eRE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eG\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e \u003cp\u003eG\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eR\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eG\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003cp\u003eG\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eImage segmentation\u003c/h3\u003e\n\u003cp\u003eImage segmentation is a critical methodology in identifying WCLWD using UAV multispectral images. This technique increases the accuracy of disease identification as it involves performing region of interest based on spectral signatures, which represent plant health and stress (Mart\u0026iacute;nez-Casasnovas et al. \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Mia et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Object-based image classification enables the contextual information, improving classification accuracy compared to that of the pixel-based approach (Ruwaimana et al. \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; De Castro et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e\n\u003ch3\u003eImage Classification\u003c/h3\u003e\n\u003cp\u003e. This approach allows separation of important spectral features that need to discriminate between healthy and infected coconut palms by WCLWD. Reflection spectra of the leaves are recorded and disease symptoms on the plant associated with phytoplasma infection, such as chlorosis and necrosis are identified based on specific wavelengths (De Silva et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2021\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The model is trained on classified data and is able to make correlations with healthy and diseased palms.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eModel Evaluation\u003c/h2\u003e \u003cp\u003eAs a performance indicator, the confusion matrix generated using the ENVI software is used in the method for the assessment of a classification accuracy of the model for detection of WCLWD, based on the object based image classification of UAV multispectral images. The classification of the test results can be observed by a confusion matrix containing actual positive, actual negative, predicted positive, and predicted negative and these are useful to calculate several performance measures: They include accuracy, precision, recall/sensitivity, and specificity (Hejmanowska et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Riehl et al. \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). This will mean that the accuracy of the classified images can be compared with ground truth to obtain a measure of the classification accuracy (Heydarian et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Moreover, the way of presenting the confusion matrix might also be improved, so that it will better be interpreted and distinguished between misclassification and the assessment of the global performance of the model (Garc\u0026iacute;a-Balboa et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Luque et al. \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e provides an interpretation of the Kappa coefficient values for the classification accuracy classes (Viera and Garrett \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2005\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eKappa coefficient description\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKappa coefficient\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDescription\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eLess than chance agreement\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.01\u0026ndash;0.20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSlight agreement\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.21\u0026ndash;0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFair agreement\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.41\u0026ndash;0.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eModerate agreement\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.61\u0026ndash;0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSubstantial agreement\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0.81\u0026ndash;0.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAlmost perfect agreement\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec12\"\u003e\n \u003cp\u003eImage acquisition\u003c/p\u003e\n \u003cp\u003eThe image acquisition was performed in two flights, covering about 1.17 ha of the study area. The total flight time was about 30 minutes, and the total number of images was 1332. Figure 6 shows the flight planning, while Fig. 7 shows samples of acquired spectral images including red edge, NIR, green, RGB, red, and blue.\u003c/p\u003e\n \u003cp\u003ePre-processing Output\u003c/p\u003e\n \u003cp\u003eThe images obtained from the UAV were first corrected for their radiometric values and were georeferenced with the Rectified Skewed Orthomorphic (RSO) projection. After georeferencing, orthorectification was done to correct for skewness in the images so as to correctly represent the spatial aspect. This allows for the right alignment of every single band with geographical coordinates, making for more or less accurate analysis possible. Associated Orthorectified images were mosaiced to produce a general representation of the study area to perform the overall analysis for the multispectral data.\u003c/p\u003e\n \u003cp\u003eGround Census\u003c/p\u003e\n \u003cp\u003eData on occurrence and severity of WCLWD in coconut palms were collected by ground census in the study area. The census was carried out in association with the Coconut Research Institute (CRI) of Sri Lanka, which supplied the field equipment and technical expertise. In the study area, the census involved individual visits to every coconut palm for visual inspection and laborory testing for symptoms of WCLWD, such as yellowing, wilting, and necrosis leaves, stem bleeding and rotting. Yellow labels were applied to the diseased palms and their coordinates noted using a handheld GPS unit. The results from the multispectral imagery analysis were validated using ground census data; predictive models for WCLWD using object based image analysis (OBIA) were developed. Figure 8 is the ground census data plotted with ArcMap software and digitised.\u003c/p\u003e\n \u003cp\u003eSegmentation\u003c/p\u003e\n \u003cp\u003eThese pre-processed images were processed by the OBIA applying 5 single bands and 26 band combinations. Each image combination was processed with a watershed segmentation using ENVI 5.0 software. To find out the best parameters related to the Segment and Merging processes, the trial and error method were conducted. The findings showed that the optimal Segment level was Edge-based at a Scale Level 50 while the Merging remained optimal at a level of 20 with the use of the FLS algorithm, this with a TKS of 3. These values were selected because they allowed for distinguishing the coconut tree canopy from fronds and minimized over segmentation problems. The segmented images were then utilized for further classification employing SVM using the four categorized segmented images.\u003c/p\u003e\n \u003cp\u003eSVM Classification\u003c/p\u003e\n \u003cp\u003eFor the SVM classification, 38 and 7 coconut palms for normal and diseased trees were selected as training samples. The training samples were randomly collected and well covered across the study area. SVM was used instead of other classifiers due to various studies calling for a better classification accuracy, particularly for multispectral images (Pal and Mather 2005; Ballanti et al. 2016).\u003c/p\u003e\n \u003cp\u003eThe SVM classification results were illustrated in pseudo-color images with green color representing unaffected trees and red color for diseased trees. The SVM classifiers were optimized using a trial-and-error method whereby the optimum classification accuracy was achieved. The parameters utilized in the current research were kernel type as Radial Basis Function (RBF) kernel type and gamma in kernel function as 0.333 with 100 parameters being imposed to Germ. The Degree of Kernel Polynomial (DoKP) was adjusted at 1 while the Bias in Kernel Function (BiKF) was also adjusted at 1. The converging value was defined as 5, which equals a 95% confidence level for classification purposes. The analysis also put the unclassified in another class so as to reduce the level of error in the eventual classification. The accuracy of the classification outputs was evaluated by comparing the OBIA output with the ground census and calculating the percentage of accuracy and kappa coefficient from the confusion matrix. Table 3 illustrates the accuracy rankings and the corresponding Kappa coefficients for each combination.\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 3\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eAccuracy Assessment of Classification Output for Individual Band and Different Band Combinations.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eBand combinations\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eRank\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOverall accuracy (% )\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eKappa coefficient\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e65.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3453\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eG\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e26\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e67.19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3466\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e67.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.377\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e71.89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.407\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e53.88\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.1194\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e72.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.42\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;R\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e74.43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.43\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e76.11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.45\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e68.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.3\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eG\u0026thinsp;+\u0026thinsp;R\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e55.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.1282\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eG\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e74.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.44\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eG\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75.34\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4512\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eR\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e59.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.225\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eR\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e76.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4725\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eRE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e70.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.35\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e78.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e79.04\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.49\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e77.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.475\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e76.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e76.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.455\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.44\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eG\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4788\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eG\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e78.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4838\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eR\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e71.67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4109\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eG\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e79.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.4706\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e78.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.485\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e78.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.48\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e79.25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.493\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e76.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eG\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e75.75\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.445\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eB\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e77.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.485\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study effectively illustrates the impact of various band combinations on classification accuracy. The integration of diverse spectral bands leads to significant accuracy enhancement, evidenced by the combination of B\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR, which has the highest overall accuracy of 79.25% and a Kappa coefficient of 0.493, than other combinations. Comparatively, the simplest band combinations, such as individual bands or pairs like B and NIR, yielded lower accuracy, with B alone at 65.03% and NIR at 53.88%. When comparing three-band combinations B\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R, an accuracy of 78.0% was achieved with a Kappa coefficient of 0.48. B\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;RE, Slightly higher accuracy at 79.0%, Kappa coefficient of 0.49. G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;NIR, achieved a 78.12% accuracy with a Kappa coefficient of 0.4838. Four-band combinations generally performed better than simpler combinations. B\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;RE, Achieved an accuracy of 78.5% and a Kappa coefficient of 0.485. B\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;R\u0026thinsp;+\u0026thinsp;NIR, Higher accuracy at 78.75% with a Kappa coefficient of 0.48.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e illustrates the classification results, identifying both normal trees and those infected with WCLWD by utilizing a combination of Blue, Green, Red Edge, and Near-Infrared (B\u0026thinsp;+\u0026thinsp;G\u0026thinsp;+\u0026thinsp;RE\u0026thinsp;+\u0026thinsp;NIR) spectral bands. In Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003e, a comparative visualization of the classification results is provided, with normal trees denoted by black circles and infected trees depicted with white circles.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003ePrevious works have also reported that the SVM classifier proves to be better than other classifiers like KNN and RF for plant disease detection. For example, Chen found that when developing plant disease classifiers, the SVM classifier had precision higher than 97% while the RF was lower (Chen et al. \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The study of Rahayu F. et al. (2021) also indicates that for texture and color features, SVM yields better results compared to KNN and DT to classify the diseased leaves. These together highlight that SVM is much more accurate in plant disease detection tasks than the KNN and RF classifiers.\u003c/p\u003e \u003cp\u003eAlso Support Vector Machine (SVM) classification effectively improves the ability to classify vegetation into different classes by using its spectral traits. SVM has been employed in numerous research to distinguish the various types of plants; it has also been utilized in determining health standards Since; it is capable of analyzing numerous fields of understanding, it is effective regardless the size of the data set Essay on SVM (Buters et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe study highlights that NIR band important in remote sensing applications since the combinations with NIR produced better results throughout the assessments and demonstrated the efficiency of NIR in improving the outcomes of classification. The classification results in the near-infrared (NIR) region have been shown to increase by offering useful information about plant health and stress conditions (Hall et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The NIR band is particularly important as it is sensitive to variations in leaf chlorophyll content, which can indicate disease presence (Hu et al. \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAnalysis of the data showed that different spectral band imposed different flexibility on classification of images to changes in the biophysical characteristics of plants, in particular nitrogen and chlorophyll content. However, as studies have suggested even stronger sensitivity of the G band at canopy level than that of the R band, which rises with chlorophyll content (Fern\u0026aacute;ndez et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Instead, the R band is targeted more at chlorophyll a and b, which are very important in the photosynthesis process and can be quantified using the spectral indices (Lohner et al. \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePrevious studies have demonstrated that the potential of the RE band for physiological conditions of vegetation are consistent with studies reporting that this band is more sensitive to stress and chlorophyll content of plant canopies (Hunt et al. \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Adelabu et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Moreover, the near-infrared (NIR) band is very efficient for obtaining data from moisture status and moisture content in the plant, other plant stress and biomass characterization features (Verrelst et al. \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eConsequently, we noticed the highest accuracy of each entire band in the B band 65.01%, RE band 71.89%, G band 67.19%, R band 67.13%, and NIR band 53.88%. Our findings also revealed that stress on plants and the reduction in chlorophyll a and b, as indicated by the RE and R bands were critical factors that influenced disease condition, seconded by plant/leaf nitrogen and pigment loss as indicated by the G band.\u003c/p\u003e \u003cp\u003eBased on an integration of NIR and RE bands, great capacity with the disease diagnosis indications processing has been established. The concurrence of these bands has been found to increase differentiation of disease severity of vegetative signs, increasing diagnostic effectiveness (Heim et al. \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Liu et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). In particular, the application of multispectral imaging has been discovered to provide positive outcomes in disease identification in agricultural crops since spectral indices were used successfully to detect disease symptoms in various crops (Mahlein et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). In addition, UAV-based multispectral images are also useful for disease diagnosis in agriculture and provide a basis for effective plant health diagnosis and disease diagnosis in agriculture, which provides a good basis for a combination of NIR and RE bands for better identification (Liang et al. \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Tait et al. \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eNainanayake et al. (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2016a\u003c/span\u003e) created a modified vegetation index including the R, G, and twice B bands and developed an indicative technique that could improve the identification of coconut tree disease with an accuracy of more than 80%. Meanwhile, (Xavier et al. \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) identified four stages of Ramulia leaf blight disease of cotton using G, R, and NIR bands of the image separately. It can differentiate early, severe conditions, stated they, but it can\u0026rsquo;t be used to differentiate healthy from severe, and healthy from mild.\u003c/p\u003e \u003cp\u003eThe difficulty in discriminating trees from the ground is the most challenging undertaking experienced in this study, especially in very dense vegetated areas. Because of the canopy structure of complex canopies and of foliage overlapping, it is hard to accurately classify trees individually. These issues can be overcome using object based image analysis (OBIA) techniques by segmenting the images into meaningfully segmented objects rather than performing pixel based classifications. The use of this approach helps enhances classification outcomes by providing an understanding of the spatial relationships between trees and their surroundings (Buters et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFurthermore, the presence of other diseases, environmental stresses, and pest infestations, as well as all the other factors known to affect crop productivity tends to affect the reflectance values selected from the multispectral images, hence compromising the accuracy of the classifications. Reflectance values depend on a number of biotic and abiotic factors that can cause similarity between the symptoms of different diseases. For instance, pest damage leaves\u0026rsquo; reflectance and environmental stresses such as drought and nutrient availability can distort reflectance characteristics, which makes it hard to differentiate between healthy or diseased plants only from the spectral data (Abdulridha et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zhang et al. \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) .\u003c/p\u003e \u003cp\u003eFurther research should be done to overcome the difficulties in distinguishing symptom overlap from those of other diseases or pests. This could involve the study of hyperspectral imaging, which offers a wider spectrum, that is, it probably covers a wider spectral region resolution, which could help distinguish plant health status. Additionally, the utilization of multispectral data in conjunction with other more higher level of machine learning algorithms can lead to an even better identification of disease present in coconut trees (Loladze et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn this study detect WCLWD-infected palms, the multispectral images of the plantation obtained with the use of the UAV were subjected to OBIA analysis, including segmentation and merging with the SVM classifier. Thus, the results indicated that different band combinations and individual band accuracy varied, highlighting the importance of selecting the optimal combination for accurate analysis. It was moderate and ranged from 53.88 to 71.89 regarding the accuracy of the individuals\u0026rsquo; band\u0026rsquo;s performances. Conversely, band image combinations of more than three bands. Further confirming the findings of other researchers, with the maximum accuracy 79.25%, arising from B _ G _ RE _NIR. This shows that OBIA analysis with a four-band combination of multispectral images can be used to detect coconut trees that are infected with WCLWD. Accurate classification and interpretation of the data is still a challenge due to variation in environmental factors and confusing signs similar to other coconut diseases. Therefore, this study recommends that a combination of molecular biology, genetics, and other relevant imaging techniques is useful and needed to further design the management techniques of WCLWD. Further study employing an adequate OBIA analysis experiment that incorporates sophisticated classifiers like support vector machine (SVM) and random forest (RF) alongside a complex template match algorithm while trying to detect early WCLWD infection in coconut palms using multispectral band combinations and hyperspectral images should be done. Explore temporal aspects of WCLWD. In what way does there occur alteration in spectral signatures due to disease state change? It is recommended that future research incorporate advanced methodologies, such as molecular biology and genetics, alongside imaging techniques to develop comprehensive management strategies for WCLWD. In terms of practical recommendations for plantation managers, it is crucial to emphasize the integration of UAV imaging with ground surveys to facilitate early detection of WCLWD. This combined approach can enhance the reliability of disease monitoring and improve management practices. Additionally, the methodologies developed for WCLWD detection could be adapted for monitoring other agricultural diseases or environmental stress factors, thereby broadening the impact of this research in precision agriculture.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eCompliance with ethical standards\u003c/h2\u003e \u003cp\u003eConflict of interest The author declares that they have no confict of interest.\u003c/p\u003e \u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAbdulridha J, Ampatzidis Y, Qureshi J, Roberts P (2020) Laboratory and UAV-Based Identification and Classification of Tomato Yellow Leaf Curl, Bacterial Spot, and Target Spot Diseases in Tomato Utilizing Hyperspectral Imaging and Machine Learning. Remote Sens 12:2732. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs12172732\u003c/span\u003e\u003cspan address=\"10.3390/rs12172732\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAdelabu S, Mutanga O, Adam E (2014) Evaluating the impact of red-edge band from Rapideye image for classifying insect defoliation levels. ISPRS J Photogramm Remote Sens 95:34\u0026ndash;41. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.isprsjprs.2014.05.013\u003c/span\u003e\u003cspan address=\"10.1016/j.isprsjprs.2014.05.013\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAhmadi P, Mansor S, Farjad B, Ghaderpour E (2022) Unmanned Aerial Vehicle (UAV)-Based Remote Sensing for Early-Stage Detection of Ganoderma. Remote Sens 14:1239. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs14051239\u003c/span\u003e\u003cspan address=\"10.3390/rs14051239\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBallanti L, Blesius L, Hines E, Kruse B (2016) Tree Species Classification Using Hyperspectral Imagery: A Comparison of Two Classifiers. Remote Sens 8:445. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs8060445\u003c/span\u003e\u003cspan address=\"10.3390/rs8060445\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarman U, Pathak C, Mazumder NK (2023) Comparative assessment of Pest damage identification of coconut plant using damage texture and color analysis. Multimed Tools Appl 82:25083\u0026ndash;25105. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11042-023-14369-2\u003c/span\u003e\u003cspan address=\"10.1007/s11042-023-14369-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eButers T, Belton D, Cross A (2019) Seed and Seedling Detection Using Unmanned Aerial Vehicles and Automated Image Classification in the Monitoring of Ecological Recovery. Drones 3:53. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/drones3030053\u003c/span\u003e\u003cspan address=\"10.3390/drones3030053\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen H, Han Y, Liu Y et al (2023) Classification models for Tobacco Mosaic Virus and Potato Virus Y using hyperspectral and machine learning techniques. Front Plant Sci 14:1211617. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpls.2023.1211617\u003c/span\u003e\u003cspan address=\"10.3389/fpls.2023.1211617\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDe Castro AI, Shi Y, Maja JM, Pe\u0026ntilde;a JM (2021) UAVs for Vegetation Monitoring: Overview and Recent Scientific Contributions. Remote Sens 13:2139. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs13112139\u003c/span\u003e\u003cspan address=\"10.3390/rs13112139\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDe Silva PHPR, Aratchige NS, Ranasinghe CS, Kantha AAFL (2021) Diseased palm removals as a strategy for the successful managementof Weligama coconut leaf wilt phytoplasma disease of coconut in SriLanka. Phytopathogenic Mollicutes 11:79\u0026ndash;85. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5958/2249-4677.2021.00013.X\u003c/span\u003e\u003cspan address=\"10.5958/2249-4677.2021.00013.X\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDe Silva PR, Perera CN, Bahder BW, Attanayake RN (2023) Nested PCR-Based Rapid Detection of Phytoplasma Leaf Wilt Disease of Coconut in Sri Lanka and Systemic Movement of the Pathogen. Pathogens 12:294. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/pathogens12020294\u003c/span\u003e\u003cspan address=\"10.3390/pathogens12020294\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDivyanth LG, Soni P, Pareek CM et al (2022) Detection of Coconut Clusters Based on Occlusion Condition Using Attention-Guided Faster R-CNN for Robotic Harvesting. Foods 11:3903. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/foods11233903\u003c/span\u003e\u003cspan address=\"10.3390/foods11233903\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFern\u0026aacute;ndez E, Gorchs G, Serrano L (2019) Use of consumer-grade cameras to assess wheat N status and grain yield. PLoS ONE 14:e0211889. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0211889\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0211889\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGani MO, Kuiry S, Das A et al (2021) Multispectral Object Detection with Deep Learning. In: Dutta P, Mandal JK, Mukhopadhyay S (eds) Computational Intelligence in Communications and Business Analytics. Springer International Publishing, Cham, pp 105\u0026ndash;117\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGarc\u0026iacute;a-Balboa JL, Alba-Fern\u0026aacute;ndez MV, Ariza-L\u0026oacute;pez FJ, Rodr\u0026iacute;guez-Avi J (2018) Analysis of Thematic Similarity Using Confusion Matrices. ISPRS Int J Geo-Inf 7:233. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/ijgi7060233\u003c/span\u003e\u003cspan address=\"10.3390/ijgi7060233\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo Y, Senthilnath J, Wu W et al (2019) Radiometric Calibration for Multispectral Camera of Different Imaging Conditions Mounted on a UAV Platform. Sustainability 11:978. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/su11040978\u003c/span\u003e\u003cspan address=\"10.3390/su11040978\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHall O, Dahlin S, Marstorp H et al (2018) Classification of Maize in Complex Smallholder Farming Systems Using UAV Imagery. Drones 2:22. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/drones2030022\u003c/span\u003e\u003cspan address=\"10.3390/drones2030022\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHeim R, Wright I, Scarth P et al (2019) Multispectral, Aerial Disease Detection for Myrtle Rust (Austropuccinia psidii) on a Lemon Myrtle Plantation. Drones 3:25. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/drones3010025\u003c/span\u003e\u003cspan address=\"10.3390/drones3010025\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHejmanowska B, Kramarczyk P, Głowienka E, Mikrut S (2021) Reliable Crops Classification Using Limited Number of Sentinel-2 and Sentinel-1 Images. Remote Sens 13:3176. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs13163176\u003c/span\u003e\u003cspan address=\"10.3390/rs13163176\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHeydarian M, Doyle TE, Samavi R (2022) MLCM: Multi-Label Confusion Matrix. IEEE Access 10:19083\u0026ndash;19095. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ACCESS.2022.3151048\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2022.3151048\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu X, Yang L, Zhang Z (2020) Non-destructive identification of single hard seed via multispectral imaging analysis in six legume species. Plant Methods 16:116. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13007-020-00659-5\u003c/span\u003e\u003cspan address=\"10.1186/s13007-020-00659-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHunt ER, Daughtry CST, Eitel JUH, Long DS (2011) Remote Sensing Leaf Chlorophyll Content Using a Visible Band Index. Agron J 103:1090\u0026ndash;1099. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2134/agronj2010.0395\u003c/span\u003e\u003cspan address=\"10.2134/agronj2010.0395\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKanatiwela-de Silva C, Damayanthi M, De Silva N et al (2019) Immunological detection of the Weligama coconut leaf wilt disease associated phytoplasma: Development and validation of a polyclonal antibody based indirect ELISA. PLoS ONE 14:e0214983. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0214983\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0214983\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumara ADNT, Chandrashekharaiah M, Kandakoor SB, Chakravarthy AK (2015) Status and Management of Three Major Insect Pests of Coconut in the Tropics and Subtropics. In: Chakravarthy AK (ed) New Horizons in Insect Science: Towards Sustainable Pest Management. Springer India, New Delhi, pp 359\u0026ndash;381\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiaghat S, Ehsani R, Mansor S et al (2014) Early detection of basal stem rot disease (\u003cem\u003eGanoderma\u003c/em\u003e) in oil palms based on hyperspectral reflectance data using pattern recognition algorithms. Int J Remote Sens 35:3427\u0026ndash;3439. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/01431161.2014.903353\u003c/span\u003e\u003cspan address=\"10.1080/01431161.2014.903353\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiang W, Zhang H, Zhang G, Cao H (2019) Rice Blast Disease Recognition Using a Deep Convolutional Neural Network. Sci Rep 9:2869. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-019-38966-0\u003c/span\u003e\u003cspan address=\"10.1038/s41598-019-38966-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu J, Xiang J, Jin Y et al (2021) Boost Precision Agriculture with Unmanned Aerial Vehicle Remote Sensing and Edge Intelligence: A Survey. Remote Sens 13:4387. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs13214387\u003c/span\u003e\u003cspan address=\"10.3390/rs13214387\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLohner SA, Biegert K, Hohmann A et al (2022) Chlorophyll- and anthocyanin-rich cell organelles affect light scattering in apple skin. Photochem Photobiol Sci 21:261\u0026ndash;273. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s43630-021-00164-1\u003c/span\u003e\u003cspan address=\"10.1007/s43630-021-00164-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLoladze A, Rodrigues FA, Toledo F et al (2019) Application of Remote Sensing for Phenotyping Tar Spot Complex Resistance in Maize. Front Plant Sci 10:552. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fpls.2019.00552\u003c/span\u003e\u003cspan address=\"10.3389/fpls.2019.00552\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuque A, Mazzoleni M, Carrasco A, Ferramosca A (2022) Visualizing Classification Results: Confusion Star and Confusion Gear. IEEE Access 10:1659\u0026ndash;1677. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ACCESS.2021.3137630\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2021.3137630\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMahlein A-K, Steiner U, Dehne H-W, Oerke E-C (2010) Spectral signatures of sugar beet leaves for the detection and differentiation of diseases. Precis Agric 11:413\u0026ndash;431. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11119-010-9180-7\u003c/span\u003e\u003cspan address=\"10.1007/s11119-010-9180-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMart\u0026iacute;nez-Casasnovas JA, Sandon\u0026iacute;s-Pozo L, Escol\u0026agrave; A et al (2021) Delineation of Management Zones in Hedgerow Almond Orchards Based on Vegetation Indices from UAV Images Validated by LiDAR-Derived Canopy Parameters. Agronomy 12:102. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/agronomy12010102\u003c/span\u003e\u003cspan address=\"10.3390/agronomy12010102\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMia MS, Tanabe R, Habibi LN et al (2023) Multimodal Deep Learning for Rice Yield Prediction Using UAV-Based Multispectral Imagery and Weather Data. Remote Sens 15:2511. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs15102511\u003c/span\u003e\u003cspan address=\"10.3390/rs15102511\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNainanayake AD, Gunathilake J, Kumarathunga MDP et al (2016a) Limitation in the use of spectral analysis to detect Weligama coconut leaf wilt disease affected palms in Southern Sri Lanka. COCOS 22:13\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4038/cocos.v22i1.5808\u003c/span\u003e\u003cspan address=\"10.4038/cocos.v22i1.5808\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNainanayake AD, Kumarathunga MDP, De Silva PHPR (2016b) A survey of land for Weligama coconut leaf wilt disease affected palms outside the declared boundary in the Southern Province. COCOS 22:57\u0026ndash;64. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4038/cocos.v22i1.5812\u003c/span\u003e\u003cspan address=\"10.4038/cocos.v22i1.5812\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePal M, Mather PM (2005) Support vector machines for classification in remote sensing. Int J Remote Sens 26:1007\u0026ndash;1011. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/01431160512331314083\u003c/span\u003e\u003cspan address=\"10.1080/01431160512331314083\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePatr\u0026iacute;cio DI, Rieder R (2018) Computer vision and artificial intelligence in precision agriculture for grain crops: A systematic review. Comput Electron Agric 153:69\u0026ndash;81. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compag.2018.08.001\u003c/span\u003e\u003cspan address=\"10.1016/j.compag.2018.08.001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerera SACN, Herath HMNB, Wijesekera HTR et al (2016a) Evaluation of coconut germplasm in Weligama and Matara area of the Southern Province of Sri Lanka for resistance to Weligama coconut leaf wilt disease. COCOS 21:15\u0026ndash;20. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4038/cocos.v21i0.5803\u003c/span\u003e\u003cspan address=\"10.4038/cocos.v21i0.5803\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerera SACN, Kumarasinghe WM, Gunasekara TMCP (2016b) Assessing the performance of fruit colour based phenotypes of tall (Typica) coconuts (Cocos nucifera L.) in Sri Lanka. COCOS 22:25\u0026ndash;29. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4038/cocos.v22i1.5809\u003c/span\u003e\u003cspan address=\"10.4038/cocos.v22i1.5809\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrades A, Salum UN, Pioch D (2016) New era for the coconut sector. What prospects for research? OCL 23:D607. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1051/ocl/2016048\u003c/span\u003e\u003cspan address=\"10.1051/ocl/2016048\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRahayu FBR, Mudjirahardjo P, Muslim MA (2021) Leaf Diseases Classification on Peanut Leaves Based on Texture and Colour Features. Int J Comput Appl Technol Res 10:149\u0026ndash;155. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7753/IJCATR1006.1004\u003c/span\u003e\u003cspan address=\"10.7753/IJCATR1006.1004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRajan P (2011) Gradient Outbreak of Coconut Slug Caterpillar, Macroplectra nararia Moore in East Coast of India. CORD 27:7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.37833/cord.v27i1.124\u003c/span\u003e\u003cspan address=\"10.37833/cord.v27i1.124\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRiehl K, Neunteufel M, Hemberg M (2023) Hierarchical confusion matrix for classification performance evaluation. J R Stat Soc Ser C Appl Stat 72:1394\u0026ndash;1412. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1093/jrsssc/qlad057\u003c/span\u003e\u003cspan address=\"10.1093/jrsssc/qlad057\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRodr\u0026iacute;guez AC, Daudt RC, D\u0026rsquo;Aronco S et al (2021) Robust Damage Estimation of Typhoon Goni on Coconut Crops with Sentinel-2 Imagery. Remote Sens 13:4302. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs13214302\u003c/span\u003e\u003cspan address=\"10.3390/rs13214302\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRuwaimana M, Satyanarayana B, Otero V et al (2018) The advantages of using drones over space-borne imagery in the mapping of mangrove forests. PLoS ONE 13:e0200288. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1371/journal.pone.0200288\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0200288\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSubajiny S, Dilini B, Terrence M (2018) A comparative study on stability of different types of coconut (Cocos nucifera) oil against autoxidation and photo-oxidation. Afr J Food Sci 12:216\u0026ndash;229. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5897/AJFS2018.1695\u003c/span\u003e\u003cspan address=\"10.5897/AJFS2018.1695\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTait L, Bind J, Charan-Dixon H et al (2019) Unmanned Aerial Vehicles (UAVs) for Monitoring Macroalgal Biodiversity: Comparison of RGB and Multispectral Imaging Sensors for Biodiversity Assessments. Remote Sens 11:2332. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs11192332\u003c/span\u003e\u003cspan address=\"10.3390/rs11192332\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTu Y-H, Johansen K, Phinn S, Robson A (2019) Measuring Canopy Structure and Condition Using Multi-Spectral UAS Imagery in a Horticultural Environment. Remote Sens 11:269. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs11030269\u003c/span\u003e\u003cspan address=\"10.3390/rs11030269\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVerrelst J, Rivera JP, Gitelson A et al (2016) Spectral band selection for vegetation properties retrieval using Gaussian processes regression. Int J Appl Earth Obs Geoinf 52:554\u0026ndash;567. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jag.2016.07.016\u003c/span\u003e\u003cspan address=\"10.1016/j.jag.2016.07.016\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eViera A, Garrett J (2005) Understanding Interobserver Agreement: The Kappa Statistic. Fam Med 37:360\u0026ndash;363\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWijesekara HTR, Perera SACN, Bandupriya D et al (2020) Detection of Weligama Coconut Leaf Wilt Disease Phytoplasma by Real-Time Polymerase Chain Reaction. CORD 36:1\u0026ndash;5. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.37833/cord.v36i.425\u003c/span\u003e\u003cspan address=\"10.37833/cord.v36i.425\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWijesekara T, Perera L, I. W, et al (2008) Preliminary Investigation on Weligama Coconut leaf Wilt disease: A new disease in southern part of Sri Lanka\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXavier TWF, Souto RNV, Statella T et al (2019) Identification of Ramularia Leaf Blight Cotton Disease Infection Levels by Multispectral, Multiscale UAV Imagery. Drones 3:33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/drones3020033\u003c/span\u003e\u003cspan address=\"10.3390/drones3020033\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang S, Li X, Ba Y et al (2022) Banana Fusarium Wilt Disease Detection by Supervised and Unsupervised Methods from UAV-Based Multispectral Imagery. Remote Sens 14:1231. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs14051231\u003c/span\u003e\u003cspan address=\"10.3390/rs14051231\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-plant-diseases-and-protection","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jpdp","sideBox":"Learn more about [Journal of Plant Diseases and Protection](https://www.springer.com/journal/41348)","snPcode":"41348","submissionUrl":"https://www.editorialmanager.com/jpdp","title":"Journal of Plant Diseases and Protection","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Agricultural disease monitoring, Coconut disease detection, Object-based image analysis (OBIA), Remote sensing, Unmanned aerial vehicle (UAV), Weligama Coconut Leaf Wilt Disease (WCLWD)","lastPublishedDoi":"10.21203/rs.3.rs-5767642/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5767642/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eWeligama Coconut Leaf Wilt Disease (WCLWD), a major threat to the coconut industry in Sri Lanka, has resulted in large economic losses (reduced productivity and high mortality rate) among infected palm. Early diagnosis is challenging and unreliable due to the low sensitivity of conventional disease detection methods like visual inspections and laboratory testing. In order to overcome these constraints, this study used object-based image analysis (OBIA) in combined with multispectral imaging using an unmanned aerial vehicle (UAV) to identify and categorize WCLWD in coconut palms. To differentiate between healthy and infected trees, Support Vector Machine (SVM) classification was used to analyze UAV images taken in five spectral bands: red, green, blue, red edge, and near infrared. The four band combination of 'blue', 'green', 'red-edge' and 'near infrared' was found to be the best of those tested, with an accuracy of 79.25% and a moderate agreement, based on the kappa coefficient of 0.493. The accuracy of this was then validated against a field survey ground truth data. Results show that overland biomass detection using OBIA methods with UAV multispectral imaging offers a feasible means to identify WCLWD, but that further classifier work and extra sources of data can improve accuracy. Results show the possibility of advanced remote sensing technologies for improve the detection of coconut WCLWD and support for managing the spread of disease in coconut plantations.\u003c/p\u003e","manuscriptTitle":"Detecting Weligama Coconut Leaf Wilt Disease in Coconut Using UAV-Based Multispectral Imaging and Object-Based Classification","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-02-04 08:55:45","doi":"10.21203/rs.3.rs-5767642/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"","date":"2025-01-31T12:18:48+00:00","index":0,"fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-01-29T09:51:12+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"Journal of Plant Diseases and Protection","date":"2025-01-21T09:39:39+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-01-20T13:37:52+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Plant Diseases and Protection","date":"2025-01-09T20:08:21+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-plant-diseases-and-protection","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"jpdp","sideBox":"Learn more about [Journal of Plant Diseases and Protection](https://www.springer.com/journal/41348)","snPcode":"41348","submissionUrl":"https://www.editorialmanager.com/jpdp","title":"Journal of Plant Diseases and Protection","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"3fa6889c-28df-4851-9c76-8b352ab32a9a","owner":[],"postedDate":"February 4th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-07-14T16:03:57+00:00","versionOfRecord":{"articleIdentity":"rs-5767642","link":"https://doi.org/10.1007/s41348-025-01115-z","journal":{"identity":"journal-of-plant-diseases-and-protection","isVorOnly":false,"title":"Journal of Plant Diseases and Protection"},"publishedOn":"2025-07-12 15:57:03","publishedOnDateReadable":"July 12th, 2025"},"versionCreatedAt":"2025-02-04 08:55:45","video":"","vorDoi":"10.1007/s41348-025-01115-z","vorDoiUrl":"https://doi.org/10.1007/s41348-025-01115-z","workflowStages":[]},"version":"v1","identity":"rs-5767642","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5767642","identity":"rs-5767642","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}