Pavement Surface Condition Rating of Flexible Pavement based on Artificial Intelligence based Deep Learning Technique

Pavement Surface Condition Rating of Flexible Pavement based on Artificial Intelligence based Deep Learning Technique | 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 Article Pavement Surface Condition Rating of Flexible Pavement based on Artificial Intelligence based Deep Learning Technique Athiappan K, Chitra Devi R, J Naveenkumar, C Makendran, Mehdi Gheisari, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9207586/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Developing countries were struggling to maintain its road networks in fit surface conditions due to adherence of the conventional method of evaluating the pavement surface condition using multiple physical equipment. The conventional method to measure the multiple distresses is a time consuming and costlier process. Hence in this research, we propose a deep learning-based Pavement Surface Condition Rating of plastic roads. The proposed hybrid model was adopted to assess the pavement surface condition assessment aligned to Pavement condition rating recommended by IRC 82:2015 used in pavement maintenance of Indian roads. In this research, 4k video image of pavement having distresses such as crack, potholes, patches and raveling are acquired by unmanned aerial vehicle mounted with camera. The collected video dataset was used to develop distress prediction model using YOLOv8 integrated with Gray-Level Co-occurrence Matrix (GLCM) and TAMURA feature extraction algorithm. It is observed from the results that the pavement condition rating based on conventional and deep learning- based Image processing technique has high degree of correlation. Physical sciences/Engineering Physical sciences/Mathematics and computing Flexible Pavement Pavement Condition Rating (PCR) AI Model Distress Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 1. Introduction India is the second largest road network with the length of 6,617,900 km including all types of roads based on classification of construction catering 85% of total passenger traffic and 71% of the total fright traffic. Among the total length, 4,102,000 km roads are surfaced with 2% of concrete pavements and 98% of bituminous pavements[1]. Even though the initial construction cost of bituminous pavement is low but need higher frequency maintenance due to periodic deformation and cracking due to viscoelastic nature of bitumen used adhesive in the bituminous bitumen[2]. With proper maintenance of bituminous pavement, it can sustain its complete life cycle of 20 to 30 years, else the pavement will lose its surface and structural integrity before its life span, nearly within 8 to 12 years[3]. India has constructed the low volume rural roads length of 56,875 km with plastic incorporated bituminous pavement in view of reducing the frequency of maintenance, use of conventional bitumen and environmental impact under PMGSY scheme [4]. India is allocating nearly 1% of overall highway development fund for the highway maintenance. every year. Due to poor pavement maintenance and management, nearly 9,109 fatal accident death on road due to potholes alone in India for the period of 2019 to 2023 period[5]. Most of the roads on India subjected to recurring surface distress due to lack of proper drainage and implementation enforcement on restricting higher axle loads than the standard axle loads permitted[6]. Even though India has standardized the maintenance for roads through IRC and MORTH specification but the adaptation level of maintenance for national highways are higher and very low in case of low volume rural roads due to scarcity of enough fund allotted for the highway maintenance [[7,8]. Nearly 65% of the India total network was shared by the rural low volume roads, lack of proper maintenance will lead to faster deterioration of pavement and poor access to small villages and towns (Vandana Tare 2013, Randhawa et al. 2025). India is relying on convention methods for pavement surface evaluation such as visual condition survey with disadvantage subjective, inconsistent rating and lack of skilled inspector and pavement condition index (PCI) produce constructive evaluation but disadvantage of higher time consuming, scarcity of trained personnel and less cost effective [11]. Meanwhile the image processing technique was adopted for the assessment based on pavement condition Index method by taking the pantographs of various distress are found to be satisfactory [12]. In the current scenario of highway infrastructural development worldwide and in India, the conventional method of pavement surface assessment was unable to meet the demand of completing the pavement surface condition assessment to complete the pavement maintenance paved the need for advanced technique to assess the pavement surface condition [13]. In Egypt, ANN-based distress prediction model for urban roads having various surface distress are developed using various algorithm such as support vector regression, random forest and decision tree, the correlation between the observed and predicted are found to be satisfactory [14]. The performance of ANN and ML model for the prediction of pavements surface condition are compared and found out that the ML based XGBoost produced higher prediction efficiency roughness distress on pavement having thin and thick overlays[15]. Only the width of cracks and its area on the pavement surface was predicted using image processing technique using deep learning algorithm YOLO v8 algorithm has higher correlation result[16]. The pavement surface polishing was predicted by analyzing the image integrated with ML based Gray Level Co-occurrence Matrix (GLCM) algorithm and showed a promising result to predict the pavement surface texture depth condition [17]. The rut depth was predicted using the image identification algorithm and analytic hierarchy process theory, the developed model was able to predict the rut depth of with an accuracy of 0.7 [18]. AI based deepening learning assessment of the pavement condition based on the laser captured image has higher reliability but it need highly trained technical person and the seeped of data acquisition is found to be slower rate and costlier [19]. It was unable to conduct reliable AI based distress assessment due to scarcity of open-source data related to the pavement surface distress assessment on the pavement [20]. From the above literature survey, it was found out that the conventional method of pavement condition rating was time consuming, costlier and not able to meet the current pavement condition assessment. Various AI-based pavement condition assessment are only available for the foreign roads which was constructed with pavement construction material with varied physical properties and environment while comparing to the India and most of the research are based on Pavement Condition Index (PCI) which uses a scale of 1 to 100 [21]. No AI-based research are available related to Indian context using Pavement Condition Rating (PCR) used for the pavement surface condition assessment which uses a scale of 1 to 3 according to IRC 82 2015. To overcome this issue, we proposed a deep learning-based pavement surface condition rating of plastic roads on south Indian roads in this research work. 2. Methodology In this research, the flexible pavement distress is determined based on the conventional method recommended by the IRC 82: 2015 and compared with the prediction model developed based on deep learning-based image processing technique. The study area selected for this research are 2000m length of plastic road constructed in Madurai corporation limit. Madurai, Tamil Nadu, India. The age of road was found to be 9 years and falls under traffic condition of low volume roads according to IRC 37: 2018 classification. In the conventional method of pavement surface evaluation, the distress such as cracking, potholes, ravelling and patches found on the above study area was determined using the manual measuring equipment tape to compute the area of distress and converted to weighted rating value for each distress. Each weighted rating value was summed to get final rating value and the condition of pavement is determined. In the artificial Intelligence based image processing technique, the video image acquisition of the entire road stretch considered for study was captured using the UAV (drones) fitted with high resolution camera. The data collected as video image was pre-processed and subjected to develop deep- learning model used to classify the pavement surface defects cracks, patches, potholes and ravelling. The results from the manual methods and image processing technique are compared and validated for the real time usage. The schematic representation of methodology used in this research was shown in Fig. 1 . 3. Pavement Surface Condition Rating by Conventional method The rating and conditions of the corresponding distress found in the study stretch such as cracking, ravelling, potholes and patches are determined by area of each distress with respect to total area of the pavement under surface condition evaluation by manual methods using the corresponding appropriate equipment’s. The irregular area distress was calculated using the Simpson rule. The ratings and conditions of each distress range with respect to range of individual distress according to IRC 82-2015 are given in Table 1 . The multiplying factor used for the individual distress for calculating weighted rating value according to the IRC 82:2015 was given in the Table 2 . The type of distress found on the study stretch was shown in Fig. 1 . The estimated ratings and conditions of induvial distress and overall condition of pavement considered for research are shown in Table 3 . From the table the overall rating and condition of the study stretch (pavement) was found to be 1.765 and fair condition respectively. Table 1 Pavement Condition and rating based on range of individual distress according to IRC 82-2015 Types of distress or Defects Range of Distress or defects Cracking% > 15 5–15 10 5–10 0.5 > 0 and 10 1–10 < 1 Rating 1 1.1-2 2.1-3 Conditions Poor Fair Good Table 2 Multiplying factor for calculating Weighted Rating Value Table according to IRC 82-2015 S.No Pavement distress or defects Multiplier factor 1. Cracking 1.0 2. Patches 0.75 3 Potholes 0.5 4. Ravelling 0.75 Table 3 Estimated ratings and conditions of individual distress and overall condition of pavement considered for study Land Mark Total length measured (m) Types of distress found Area of distress(m 2 ) PCI density (%) Rating based on IRC 82:2015 Condition (based on IRC 82:2015) Main parking 400 Patches 40.430 1.55 1.2 Fair Potholes 1.687 0.06 1.1 Fair B -halls 450 Patches 8.244 2.11 1.3 Fair Potholes 0.636 0.162 1.1 Fair Back gate to arch dept 301 Patches 12.818 1.216 1.2 Fair Potholes 2.545 0.241 1.3 Fair Back gate to front gate 840 Patches 18.085 2.15 1.3 Fair Potholes 3.986 0.474 1.8 Fair Cracks 0.722 0.8 3 Good Ravelling 13.88 1.65 2.9 Good Table 4 Estimated Rating of study stretches (pavement) based on percentage of overall distress Distress type Percentage of distress (%) Rating of Individual distress Multiplying factor Weighted rating value Cracking 0.8 3 1.0 3.0 Patches 7.026 1.7 0.75 1.275 Potholes 0.937 1.2 0.5 0.6 Ravelling 1.65 2.9 0.75 2.175 Final Rating Value 1.765 Condition Fair 4. Deep Learning-based pavement condition rating The deep learning-based approach was proposed to detect the pavement surface defects. An unmanned aerial vehicle with high-resolution camera was used to acquire the high-quality 4k video image of the pavement surface. The designed hardware setup can works optimal under any weather conditions i.e. upto 50 o C and 85% humidity. The 4k video was then preprocessed by splitting into individual image frames subjected to annotation and augmentation process. The main intention of this phase is to filter out the blur images, resize each of the images to 640x640 so as to support the deep learning models and also to reduce the computation time involved in the training period. In addition, the brightness of the image was improved to get better visibility on the distress. The processing carried out in this stage helps the models to learn effectively, and thereby save time with clean and balanced data. The core part of this proposed work is to use the hybrid model for the pavement condition analysis. The YOLOv8 was used to extract the deep visual features of the pavement, GLCM detects the surface irregularities and Tamura algorithm extracts the textual features. GLCM is used for the texture analysis of the dataset based on pixel density in the greyscale images. GLCM identifies the pavement surface defects based on statistical features of dataset (images) such as contrast, correlation, energy reflected, dissimilarity and entropy of the image. TAMURA feature extraction mimics the human texture perception that has the ability to detect the pavement surface irregularities. It classifies the defects based on the following parameters such as coarseness, contrast, directionality, line-likeness, regularity and roughness. The texture features extracted from GLCM and Tamura methods are combined into a comprehensive feature vector for each road surface image. YOLOv8 architecture consist of backbone network, model neck and detection head. The essential features from the input images are extracted using the model C2fbackbone structure. The features extracted by the backbone using multiple layers are refined and enhanced at the model neck using path aggregation network used to detect small pavement surface defects like cracks. The detection head is used to predict the bounding box, various defect label and confidence scores and aids in detect the pavement surface defects with high accuracy. Finally, the user interface was implemented to depict the surface defects on the pavement by using pavement condition rating according to IRC 82-2015. 4.1 Dataset Collection and Pre-processing The dataset of the entire study stretch was collected using the unmanned aerial vehicle (UAV) fitted with the high-resolution video camera with capability of capturing video frame of 4k (7680 X 4320 pixels) at the rate of 60 frames per second having a video format of MOV (H.264/MPEG-4 compression). The UAV can operate at a temperature of 0⁰C to 50⁰C up to 85% of humidity. It is bundled with AF-P DX NIKKOR 18-55mm f/3.5-5.6G VR lens. The datasets were collected in different lighting, environmental factors such as different temperature, angles, distances and illuminance (from morning 7am to evening 5pm at an interval of 1 hour). The data pre-processing involves converting the obtained image into standardized image of 640X640 pixels compatible to computer vision model YOLOv8. It detects the defects in the pavement surface by breaking the video into individual done by image frame. The redundant or blurry frames are removed to avoid duplication followed by improving contrast, brightness, and sharpness and annotation prepared using Roboflow. The annotated images are subjected to the different types of image augmentation such as geometric transformation, photometric transformation, noise-based augmentation. To ensure accurate localization of defects, correct label alignment and reliable training data, the bounding boxes of annotated image box are automatically adjusted to match the transformed images. The annotated images with defects and without defects are shown in Fig. 3 a & 3 b respectively. 4.2 Classification and Labelling of Dataset The dataset collected were divided into three categories of usage, 70 percentage used for model training, 20 percentage used for model validation and 10 percentage of data used for the model testing. The data used for the CNN training model are classified into multiple classes. The positive class image represents the image with road surface defects (potholes, patches, ravelling and cracks) and negative class image represents the road without surface defects. The 70% of the image dataset was used to train the CNN model for learning the presence and absence of defects. The trained model is used to predict the presence and absence of defects in new or unseen images that are shown in Fig. 4 and Fig. 5 respectively. The multiple class dataset includes images of four different types of defects such as cracks, patches, potholes and ravelling that are observed in this research. 4.3 Pavement Surface Defect - Model Training The YOLOv8 architecture was initialized and trained with 50 epochs where 8 images per batch were considered to avoid underfitting and memorizing the training data. The multiclass classification was done using cross-entropy loss, bounding box regression loss and abjectness or confidence loss. The Adaptive Moment Estimation (ADAM) uses momentum-based optimization and adaptive learning rates for optimizing the model parameters consisting of various sensitivities and gradient magnitudes. The training parameters such as training loss, validation loss, precision, recall, and mean Average Precision (mAP) are used to monitor the efficiency and accuracy of the training process. The trained images and the training results are shown in Fig. 6 and Fig. 7 respectively. 4.4 Pavement Surface Defect - Model Validation The Pavement Surface Defect Model developed using the deep learning network was validated using the 10% of random images from the dataset that was not utilized for training. The pavement surface defects (Potholes, patches, cracking and ravelling) are provided as input and the pavement surface defects detection ability was assessed as shown in Fig. 8 . The validation results observed for the proposed model has a precision of 72.3%, recall score of 68.1%. The confusion matrix for the true and predicted pavement surface defects are shown in Fig. 9. The precision confidence curve value varies from 1.0 to 0.658 as shown in Fig. 10 . 5. Comparative Analysis The comparison of pavement rating obtained from the conventional method and their results are compared against the deep learning-based image processing techniques. The results are depicted in the Table 5 . The variation of area for different pavement surface distress (potholes, cracking, ravelling and patches) obtained by conventional methods varies with the distress identified through deep learning-based image processing techniques. Figure 11 depicts the variation of various distress types by conventional and AI based image processing technique. Even though pavement surface detection model developed by deep learning based image processing technique has a confidence level of 1 to 0.58, the final rating value according to IRC 82: 2015 has slight variation between the conventional method and AI based image processing technique found to be 1.765 and 1.73 falls under fair condition. The huge variation of area was found for the area of potholes and ravelling detected by these two methods, whereas the area of cracking and patches detected shows less variation by both the methods. Table 5 Comparative analysis of pavement rating obtained between the conventional method and deep learning-based image processing techniques Distress type Percentage of distress by conventional method (%) Percentage of distress by AI based image processing technique (%) Rating of Individual distress by conventional method Rating of Individual distress by AI based image processing technique Multiplying factor Weighted rating value for conventional method Weighted rating value by deep learning technique Cracking 0.8 1.15 3 3 1.0 3.0 3.0 Patches 7.026 6.52 1.7 1.7 0.75 1.275 1.275 Potholes 0.48 1.73 1.2 1.0 0.5 0.6 0.5 Ravelling 1.65 4.23 2.9 2.9 0.75 2.175 2.175 Final Rating Value 1.765 1.73 Condition Fair Fair 6. Conclusion The pavement surface condition ratings value obtained based on the conventional and deep learning (DL)-based image processing techniques. For both the cases, the condition rating is found to be fair. Hence this DL-based model can be used for evaluating the pavement surface condition of roads constructed in south India using similar specification. The variation of distress estimated for the cracks and patches based on the conventional method and DL-based image processing techniques have high degree of correlation, whereas the other two distress potholes and ravelling found to have less degree of correlation. The proposed hybrid model is found to process large quantity of data necessary for modelling with higher accuracy, and have faster rate of data processing. This research work carried out is only suitable to evaluate the pavement surface conditions of the plastic roads. The study is made only on the following distress such as crack, potholes, ravelling and patches, hence the proposed model is limited to predict the mentioned distress alone. Similarly, the model is valid only in the southern Indian road conditions which reflect the bitumen material characteristics supporting the hot climatic conditions. The research can be extended to consider all types of pavement surface distress and roads all over the India with all climatic conditions and various material characteristics. Declarations Author Contribution A. wrote the manuscriptR.J.Modelling workC.Supervision of WorkR.Figure preparationEsnaashari . Grammer checkFernández-Campusano. Supervision Data Availability The datasets generated and analyzed during the current study are not publicly available due but are available from the corresponding author on reasonable request. References Ministry of Road Transport & Highways. Year End Review 2025 – Ministry of Road Transport & Highways . (2025). Guo, R. et al. Ultra-thin asphalt pavement preservation layer: Performance enhancement mechanism, key technologies, and development trends. Case Studies in Construction Materials 23 , (2025). Siva Rama Krishna, U. & Naga Satish Kumar, C. A case study on maintenance of bituminous concrete pavement considering life cycle cost analysis and carbon footprint estimation. International Journal of Construction Management 22 , 1756–1764 (2022). Ministry of Rural Development. Construction of Roads Using Plastic Wastes . (2025). PRS LEGISLATIVE RESEARCH. Demand for Grants 2024-25: Road Transport and Highways . (2024). Patil Prathisthan’, D. Y. & Jalindar, P. Effects of Bad Drainage on Roads Patil Abhijit (Corresponding author). www.iiste.orgwww.iiste.org. Nautiyal, A. & Sharma, S. Scientific approach using AHP to prioritize low volume rural roads for pavement maintenance. J. Qual. Maint. Eng. 28 , 411–429 (2022). Gertler, P. J., Gonzalez-Navarro, M., Gračner, T. & Rothenberg, A. D. Road maintenance and local economic development: Evidence from Indonesia’s highways. J. Urban Econ. 143 , (2024). Vandana Tare, H. S. G. A. B. K. M. ram. PAVEMENT DETERIORATION MODELING FOR LOW VOLUME ROADS. Journal Of The Indian Roads Congress Volume 73 , 67–891 (2013). Randhawa, S. et al. Paved or unpaved? A deep learning derived road surface global dataset from mapillary street-view imagery. ISPRS Journal of Photogrammetry and Remote Sensing 223 , 362–374 (2025). Almnam, A. & Akhmira, A. ASSESSMENT OF ROAD PAVEMENT CONDITION USING PAVEMENT CONDITION INDEX (PCI) . Shtayat, A. et al. Optimizing Road Pavement Assessment Using Advanced Image Processing Techniques. Sustainability (Switzerland) 17 , (2025). Khatoon, A. et al. Advancement in Pavement Condition Assessment: An AI-Based Computer Vision Approach. in 2024 Multimedia University Engineering Conference, MECON 2024 (Institute of Electrical and Electronics Engineers Inc., 2024). doi:10.1109/MECON62796.2024.10776158. Radwan, M. M., Faris, S. A., Barakat, A. Y. & Mousa, A. Distress-Based Pavement Condition Assessment Using Artificial Intelligence: A Case Study of Egyptian Roads. Eng 6 , (2025). Adnan, T. & Erfani, A. Explainable AI for predicting pavement roughness under maintenance and no-maintenance scenarios. Results in Engineering 29 , (2026). Ibragimov, E., Kim, Y., Lee, J. H., Cho, J. & Lee, J. J. Automated Pavement Condition Index Assessment with Deep Learning and Image Analysis: An End-to-End Approach. Sensors 24 , (2024). Fakhri, M., Pourjafar, S. V. & Daneshvari, M. H. Texture-based image analysis and explainable machine learning for polished asphalt identification in pavement condition monitoring. Sci. Rep. 15 , (2025). Zuo, D., Shao, L., Chen, B., Tan, J. & Yang, J. Pavement damage detection and evaluation based on UAV image and improved AHP. Case Studies in Construction Materials 22 , (2025). Washington DC. USING ARTIFICIAL INTELLIGENCE TO EVALUATE PAVEMENT CONDITION AND SAFETY. Federal Highway Administration 1 (2024) doi:10.21949/1521492. Saleh Abu Dabous, M. A. G. W. Z. K. H. R. A.-R. Artificial intelligence applications in pavement infrastructure damage detection with automated three-dimensional imaging- A systematic review. Alexandria Engineering Journal 117 , 510–533 (2025). IRC 82. CODE OF PRACTICE FOR MAINTENANCE OF BITUMINOUS ROAD SURFACES INDIAN ROADS CONGRESS . https://www.irc.nic.in/ (2015). Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-9207586","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":621682647,"identity":"0470bfbc-4f54-485c-877c-4aba9a85fb25","order_by":0,"name":"Athiappan K","email":"","orcid":"","institution":"JJ College of Engineering and Technology","correspondingAuthor":false,"prefix":"","firstName":"Athiappan","middleName":"","lastName":"K","suffix":""},{"id":621682648,"identity":"48fcab8c-c2b9-4f62-bcc6-cf30d0cb089e","order_by":1,"name":"Chitra Devi R","email":"","orcid":"","institution":"Dr.Sivanthi Aditanar College of Engineering","correspondingAuthor":false,"prefix":"","firstName":"Chitra","middleName":"Devi","lastName":"R","suffix":""},{"id":621682649,"identity":"39cfaaa3-9310-4700-b5d1-1e2207c1bfc3","order_by":2,"name":"J Naveenkumar","email":"","orcid":"","institution":"National Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"J","middleName":"","lastName":"Naveenkumar","suffix":""},{"id":621682650,"identity":"ddf900d0-21c8-4ec0-9c0e-9ae9cc67ca34","order_by":3,"name":"C Makendran","email":"","orcid":"","institution":"National Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"C","middleName":"","lastName":"Makendran","suffix":""},{"id":621682651,"identity":"d5457a87-aab9-49a9-98eb-1a814a166e6e","order_by":4,"name":"Mehdi Gheisari","email":"data:image/png;base64,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","orcid":"","institution":"Shaoxing University","correspondingAuthor":true,"prefix":"","firstName":"Mehdi","middleName":"","lastName":"Gheisari","suffix":""},{"id":621682652,"identity":"f1a7c978-15fa-45a2-8181-852e5164652e","order_by":5,"name":"Christian Fernández-Campusano","email":"","orcid":"","institution":"University of Santiago Chile","correspondingAuthor":false,"prefix":"","firstName":"Christian","middleName":"","lastName":"Fernández-Campusano","suffix":""},{"id":621682653,"identity":"ad80a8fa-11c6-402f-af87-caf5e5376667","order_by":6,"name":"Mehidi Esnaashari","email":"","orcid":"","institution":"K.N.Toosi University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Mehidi","middleName":"","lastName":"Esnaashari","suffix":""}],"badges":[],"createdAt":"2026-03-24 06:38:30","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9207586/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9207586/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106868960,"identity":"80507eec-3005-424c-886d-dea1b9c120b3","added_by":"auto","created_at":"2026-04-14 09:36:19","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":837983,"visible":true,"origin":"","legend":"\u003cp\u003eDistress found on the study stretches\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/0694a8948d4f9f8ba04dc10c.png"},{"id":106960771,"identity":"1271112a-7005-4cfe-aebf-123696462c46","added_by":"auto","created_at":"2026-04-15 09:23:03","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":12423,"visible":true,"origin":"","legend":"\u003cp\u003eMethodology for Deep learning and Image Processing based pavement condition evaluation\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/c359a87d53ecdaaed072e1e1.png"},{"id":106960949,"identity":"960b288a-a6ba-4e22-8079-ff1870f8f3a8","added_by":"auto","created_at":"2026-04-15 09:23:45","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":906875,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003e(a). \u003c/strong\u003eAnnotated Image with and without defects\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(b). \u003c/strong\u003eAnnotated Image with various defects (Potholes, patches, cracks and ravelling)\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/2414df4c139d8177b0869162.png"},{"id":106868962,"identity":"df4deece-921b-408a-885c-25e0cd42cd29","added_by":"auto","created_at":"2026-04-14 09:36:20","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":352886,"visible":true,"origin":"","legend":"\u003cp\u003ePositive image having road surface defects\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/f8e888c159791a658db79250.png"},{"id":106961512,"identity":"bca83454-fff4-4ecc-8b51-a71a5c6a9896","added_by":"auto","created_at":"2026-04-15 09:25:51","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":279442,"visible":true,"origin":"","legend":"\u003cp\u003eNegative image without road surface defects\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/905be57731ef5593a27602eb.png"},{"id":106868964,"identity":"e49740af-a949-4efa-8eaf-1903d7eab90e","added_by":"auto","created_at":"2026-04-14 09:36:20","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":517831,"visible":true,"origin":"","legend":"\u003cp\u003eImages Pavement Surface defects used for training\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/3c20a061ea734abd70ebdddd.png"},{"id":106961373,"identity":"d3217290-9966-4730-b139-f5cda1278c94","added_by":"auto","created_at":"2026-04-15 09:25:18","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":209399,"visible":true,"origin":"","legend":"\u003cp\u003ePavement surface defects training results\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/158bacfaac413062482d6ef4.png"},{"id":106868971,"identity":"35ce417f-950f-423c-a181-17bdc421ac34","added_by":"auto","created_at":"2026-04-14 09:36:20","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":395845,"visible":true,"origin":"","legend":"\u003cp\u003eSurface defects Validated image a) patches b) potholes\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/44758286961e61ea8d8d62d4.png"},{"id":106868968,"identity":"873602bd-3e83-4dc9-8773-c27a5667df7d","added_by":"auto","created_at":"2026-04-14 09:36:20","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":99563,"visible":true,"origin":"","legend":"\u003cp\u003eConfusion matrix for true and predicted multiclass surface defects\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/9db466e4ea6809f528c2b801.png"},{"id":106868965,"identity":"9b913eae-1e47-4156-8811-4f997a2211dd","added_by":"auto","created_at":"2026-04-14 09:36:20","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":107112,"visible":true,"origin":"","legend":"\u003cp\u003ePavement Surface defects precision- confidence curve\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/0a921c15970ed5d65171494b.png"},{"id":106868970,"identity":"7d19ebc5-9118-4ffc-984a-b5fe051619fe","added_by":"auto","created_at":"2026-04-14 09:36:20","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":15657,"visible":true,"origin":"","legend":"\u003cp\u003ePercentage variation of various distress types by conventional and AI based image processing technique\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/146d8ee3ce2a1b5c00a110df.png"},{"id":108013107,"identity":"ac60007d-8f21-4d51-9411-19470edae8a1","added_by":"auto","created_at":"2026-04-28 13:17:24","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":4549622,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9207586/v1/dd067e69-fcf1-41e1-85ff-40631b290829.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Pavement Surface Condition Rating of Flexible Pavement based on Artificial Intelligence based Deep Learning Technique","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eIndia is the second largest road network with the length of 6,617,900 km including all types of roads based on classification of construction catering 85% of total passenger traffic and 71% of the total fright traffic. Among the total length, 4,102,000 km roads are surfaced with 2% of concrete pavements and 98% of bituminous pavements[1]. Even though the initial construction cost of bituminous pavement is low but need higher frequency maintenance due to periodic deformation and cracking due to viscoelastic nature of bitumen used adhesive in the bituminous bitumen[2]. With proper maintenance of bituminous pavement, it can sustain its complete life cycle of 20 to 30 years, else the pavement will lose its surface and structural integrity before its life span, nearly within 8 to 12 years[3]. India has constructed the low volume rural roads length of 56,875 km with plastic incorporated bituminous pavement in view of reducing the frequency of maintenance, use of conventional bitumen and environmental impact under PMGSY scheme [4]. India is allocating nearly 1% of overall highway development fund for the highway maintenance. every year. Due to poor pavement maintenance and management, nearly 9,109 fatal accident death on road due to potholes alone in India for the period of 2019 to 2023 period[5]. Most of the roads on India subjected to recurring surface distress due to lack of proper drainage and implementation enforcement on restricting higher axle loads than the standard axle loads permitted[6]. Even though India has standardized the maintenance for roads through IRC and MORTH specification but the adaptation level of maintenance for national highways are higher and very low in case of low volume rural roads due to scarcity of enough fund allotted for the highway maintenance [[7,8]. Nearly 65% of the India total network was shared by the rural low volume roads, lack of proper maintenance will lead to faster deterioration of pavement and poor access to small villages and towns (Vandana Tare 2013, Randhawa et al. 2025). India is relying on convention methods for pavement surface evaluation such as visual condition survey with disadvantage subjective, inconsistent rating and lack of skilled inspector and pavement condition index (PCI) produce constructive evaluation but disadvantage of higher time consuming, scarcity of trained personnel and less cost effective [11]. Meanwhile the image processing technique was adopted for the assessment based on pavement condition Index method by taking the pantographs of various distress are found to be satisfactory [12]. In the current scenario of highway infrastructural development worldwide and in India, the conventional method of pavement surface assessment was unable to meet the demand of completing the pavement surface condition assessment to complete the pavement maintenance paved the need for advanced technique to assess the pavement surface condition [13]. In Egypt, ANN-based distress prediction model for urban roads having various surface distress are developed using various algorithm such as support vector regression, random forest and decision tree, the correlation between the observed and predicted are found to be satisfactory [14]. The performance of ANN and ML model for the prediction of pavements surface condition are compared and found out that the ML based XGBoost produced higher prediction efficiency roughness distress on pavement having thin and thick overlays[15]. Only the width of cracks and its area on the pavement surface was predicted using image processing technique using deep learning algorithm YOLO v8 algorithm has higher correlation result[16]. The pavement surface polishing was predicted by analyzing the image integrated with ML based Gray Level Co-occurrence Matrix (GLCM) algorithm and showed a promising result to predict the pavement surface texture depth condition [17]. The rut depth was predicted using the image identification algorithm and analytic hierarchy process theory, the developed model was able to predict the rut depth of with an accuracy of 0.7 [18]. AI based deepening learning assessment of the pavement condition based on the laser captured image has higher reliability but it need highly trained technical person and the seeped of data acquisition is found to be slower rate and costlier [19]. It was unable to conduct reliable AI based distress assessment due to scarcity of open-source data related to the pavement surface distress assessment on the pavement [20]. From the above literature survey, it was found out that the conventional method of pavement condition rating was time consuming, costlier and not able to meet the current pavement condition assessment. Various AI-based pavement condition assessment are only available for the foreign roads which was constructed with pavement construction material with varied physical properties and environment while comparing to the India and most of the research are based on Pavement Condition Index (PCI) which uses a scale of 1 to 100 [21]. No AI-based research are available related to Indian context using Pavement Condition Rating (PCR) used for the pavement surface condition assessment which uses a scale of 1 to 3 according to IRC 82 2015. To overcome this issue, we proposed a deep learning-based pavement surface condition rating of plastic roads on south Indian roads in this research work.\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e"},{"header":"2. Methodology","content":"\u003cp\u003eIn this research, the flexible pavement distress is determined based on the conventional method recommended by the IRC 82: 2015 and compared with the prediction model developed based on deep learning-based image processing technique. The study area selected for this research are 2000m length of plastic road constructed in Madurai corporation limit. Madurai, Tamil Nadu, India. The age of road was found to be 9 years and falls under traffic condition of low volume roads according to IRC 37: 2018 classification. In the conventional method of pavement surface evaluation, the distress such as cracking, potholes, ravelling and patches found on the above study area was determined using the manual measuring equipment tape to compute the area of distress and converted to weighted rating value for each distress. Each weighted rating value was summed to get final rating value and the condition of pavement is determined. In the artificial Intelligence based image processing technique, the video image acquisition of the entire road stretch considered for study was captured using the UAV (drones) fitted with high resolution camera. The data collected as video image was pre-processed and subjected to develop deep- learning model used to classify the pavement surface defects cracks, patches, potholes and ravelling. The results from the manual methods and image processing technique are compared and validated for the real time usage. The schematic representation of methodology used in this research was shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"3. Pavement Surface Condition Rating by Conventional method","content":"\u003cp\u003eThe rating and conditions of the corresponding distress found in the study stretch such as cracking, ravelling, potholes and patches are determined by area of each distress with respect to total area of the pavement under surface condition evaluation by manual methods using the corresponding appropriate equipment\u0026rsquo;s. The irregular area distress was calculated using the Simpson rule. The ratings and conditions of each distress range with respect to range of individual distress according to IRC 82-2015 are given in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The multiplying factor used for the individual distress for calculating weighted rating value according to the IRC 82:2015 was given in the Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. The type of distress found on the study stretch was shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The estimated ratings and conditions of induvial distress and overall condition of pavement considered for research are shown in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. From the table the overall rating and condition of the study stretch (pavement) was found to be 1.765 and fair condition respectively.\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\u003ePavement Condition and rating based on range of individual distress according to IRC 82-2015\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\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 \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTypes of distress or Defects\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eRange of Distress or defects\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCracking%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5\u0026ndash;15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRavelling%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e5\u0026ndash;10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePotholes%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;0.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;0 and \u0026lt;_0.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNIL(0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePatching%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026gt;\u0026thinsp;10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1\u0026ndash;10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRating\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.1-2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.1-3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eConditions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePoor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGood\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\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\u003eMultiplying factor for calculating Weighted Rating Value Table according to IRC 82-2015\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\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=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eS.No\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePavement distress or defects\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMultiplier factor\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCracking\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePatches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePotholes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRavelling\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEstimated ratings and conditions of individual distress and overall condition of pavement considered for study\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" 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=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLand Mark\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal length measured (m)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTypes of distress found\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eArea of distress(m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ePCI density\u003c/p\u003e \u003cp\u003e(%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eRating based on IRC 82:2015\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eCondition (based on IRC 82:2015)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eMain parking\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e400\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePatches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e40.430\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePotholes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.687\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eB -halls\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e450\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePatches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e8.244\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e2.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePotholes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.636\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.162\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eBack gate to arch dept\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e301\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePatches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e12.818\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.216\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePotholes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.545\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.241\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003eBack gate to front gate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e840\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePatches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e18.085\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e2.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePotholes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.986\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.474\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCracks\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.722\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eGood\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRavelling\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e13.88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003eGood\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEstimated Rating of study stretches (pavement) based on percentage of overall distress\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\u003eDistress type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePercentage of distress (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRating of Individual distress\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMultiplying factor\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eWeighted rating value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCracking\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePatches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.026\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.275\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePotholes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.937\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRavelling\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.175\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eFinal Rating Value\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.765\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003eCondition\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eFair\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e"},{"header":"4. Deep Learning-based pavement condition rating","content":"\u003cp\u003eThe deep learning-based approach was proposed to detect the pavement surface defects. An unmanned aerial vehicle with high-resolution camera was used to acquire the high-quality 4k video image of the pavement surface. The designed hardware setup can works optimal under any weather conditions i.e. upto 50\u003csup\u003eo\u003c/sup\u003eC and 85% humidity. The 4k video was then preprocessed by splitting into individual image frames subjected to annotation and augmentation process. The main intention of this phase is to filter out the blur images, resize each of the images to 640x640 so as to support the deep learning models and also to reduce the computation time involved in the training period. In addition, the brightness of the image was improved to get better visibility on the distress. The processing carried out in this stage helps the models to learn effectively, and thereby save time with clean and balanced data.\u003c/p\u003e\n\u003cp\u003eThe core part of this proposed work is to use the hybrid model for the pavement condition analysis. The YOLOv8 was used to extract the deep visual features of the pavement, GLCM detects the surface irregularities and Tamura algorithm extracts the textual features. GLCM is used for the texture analysis of the dataset based on pixel density in the greyscale images. GLCM identifies the pavement surface defects based on statistical features of dataset (images) such as contrast, correlation, energy reflected, dissimilarity and entropy of the image. TAMURA feature extraction mimics the human texture perception that has the ability to detect the pavement surface irregularities. It classifies the defects based on the following parameters such as coarseness, contrast, directionality, line-likeness, regularity and roughness. The texture features extracted from GLCM and Tamura methods are combined into a comprehensive feature vector for each road surface image. YOLOv8 architecture consist of backbone network, model neck and detection head. The essential features from the input images are extracted using the model C2fbackbone structure. The features extracted by the backbone using multiple layers are refined and enhanced at the model neck using path aggregation network used to detect small pavement surface defects like cracks. The detection head is used to predict the bounding box, various defect label and confidence scores and aids in detect the pavement surface defects with high accuracy. Finally, the user interface was implemented to depict the surface defects on the pavement by using pavement condition rating according to IRC 82-2015.\u003c/p\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003e4.1 Dataset Collection and Pre-processing\u003c/h2\u003e\n \u003cdiv class=\"BlockQuote\"\u003e\n \u003cp\u003eThe dataset of the entire study stretch was collected using the unmanned aerial vehicle (UAV) fitted with the high-resolution video camera with capability of capturing video frame of 4k (7680 X 4320 pixels) at the rate of 60 frames per second having a video format of MOV (H.264/MPEG-4 compression). The UAV can operate at a temperature of 0⁰C to 50⁰C up to 85% of humidity. It is bundled with AF-P DX NIKKOR 18-55mm f/3.5-5.6G VR lens. The datasets were collected in different lighting, environmental factors such as different temperature, angles, distances and illuminance (from morning 7am to evening 5pm at an interval of 1 hour). The data pre-processing involves converting the obtained image into standardized image of 640X640 pixels compatible to computer vision model YOLOv8. It detects the defects in the pavement surface by breaking the video into individual done by image frame. The redundant or blurry frames are removed to avoid duplication followed by improving contrast, brightness, and sharpness and annotation prepared using Roboflow. The annotated images are subjected to the different types of image augmentation such as geometric transformation, photometric transformation, noise-based augmentation. To ensure accurate localization of defects, correct label alignment and reliable training data, the bounding boxes of annotated image box are automatically adjusted to match the transformed images. The annotated images with defects and without defects are shown in Fig. \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e3\u003c/span\u003ea \u0026amp; \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e3\u003c/span\u003eb respectively.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\n \u003ch2\u003e4.2 Classification and Labelling of Dataset\u003c/h2\u003e\n \u003cdiv class=\"BlockQuote\"\u003e\n \u003cp\u003eThe dataset collected were divided into three categories of usage, 70 percentage used for model training, 20 percentage used for model validation and 10 percentage of data used for the model testing. The data used for the CNN training model are classified into multiple classes. The positive class image represents the image with road surface defects (potholes, patches, ravelling and cracks) and negative class image represents the road without surface defects. The 70% of the image dataset was used to train the CNN model for learning the presence and absence of defects. The trained model is used to predict the presence and absence of defects in new or unseen images that are shown in Fig. \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003e and Fig. \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e respectively. The multiple class dataset includes images of four different types of defects such as cracks, patches, potholes and ravelling that are observed in this research.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n \u003ch2\u003e4.3 Pavement Surface Defect - Model Training\u003c/h2\u003e\n \u003cp\u003eThe YOLOv8 architecture was initialized and trained with 50 epochs where 8 images per batch were considered to avoid underfitting and memorizing the training data. The multiclass classification was done using cross-entropy loss, bounding box regression loss and abjectness or confidence loss. The Adaptive Moment Estimation (ADAM) uses momentum-based optimization and adaptive learning rates for optimizing the model parameters consisting of various sensitivities and gradient magnitudes. The training parameters such as training loss, validation loss, precision, recall, and mean Average Precision (mAP) are used to monitor the efficiency and accuracy of the training process. The trained images and the training results are shown in Fig. \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e6\u003c/span\u003e and Fig. \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e7\u003c/span\u003e respectively.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003e4.4 Pavement Surface Defect - Model Validation\u003c/h2\u003e\n \u003cp\u003eThe Pavement Surface Defect Model developed using the deep learning network was validated using the 10% of random images from the dataset that was not utilized for training. The pavement surface defects (Potholes, patches, cracking and ravelling) are provided as input and the pavement surface defects detection ability was assessed as shown in Fig. \u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e8\u003c/span\u003e. The validation results observed for the proposed model has a precision of 72.3%, recall score of 68.1%. The confusion matrix for the true and predicted pavement surface defects are shown in Fig. 9. The precision confidence curve value varies from 1.0 to 0.658 as shown in Fig. \u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"5. Comparative Analysis","content":"\u003cp\u003eThe comparison of pavement rating obtained from the conventional method and their results are compared against the deep learning-based image processing techniques. The results are depicted in the Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e. The variation of area for different pavement surface distress (potholes, cracking, ravelling and patches) obtained by conventional methods varies with the distress identified through deep learning-based image processing techniques. Figure\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003e depicts the variation of various distress types by conventional and AI based image processing technique. Even though pavement surface detection model developed by deep learning based image processing technique has a confidence level of 1 to 0.58, the final rating value according to IRC 82: 2015 has slight variation between the conventional method and AI based image processing technique found to be 1.765 and 1.73 falls under fair condition. The huge variation of area was found for the area of potholes and ravelling detected by these two methods, whereas the area of cracking and patches detected shows less variation by both the methods.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparative analysis of pavement rating obtained between the conventional method and deep learning-based image processing techniques\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\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 \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDistress type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePercentage of distress by conventional method (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePercentage of distress by AI based image processing technique (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRating of Individual distress by conventional method\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eRating of Individual distress by AI based image processing technique\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eMultiplying factor\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eWeighted rating value for conventional method\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eWeighted rating value by deep learning technique\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCracking\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e1.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e3.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePatches\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.026\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6.52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1.275\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e1.275\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePotholes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e0.5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRavelling\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e2.175\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e2.175\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFinal Rating Value\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cb\u003e1.765\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u003cb\u003e1.73\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCondition\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e\u003cb\u003eFair\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u003cb\u003eFair\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"6. Conclusion","content":"\u003cp\u003eThe pavement surface condition ratings value obtained based on the conventional and deep learning (DL)-based image processing techniques. For both the cases, the condition rating is found to be fair. Hence this DL-based model can be used for evaluating the pavement surface condition of roads constructed in south India using similar specification. The variation of distress estimated for the cracks and patches based on the conventional method and DL-based image processing techniques have high degree of correlation, whereas the other two distress potholes and ravelling found to have less degree of correlation.\u003c/p\u003e \u003cp\u003eThe proposed hybrid model is found to process large quantity of data necessary for modelling with higher accuracy, and have faster rate of data processing. This research work carried out is only suitable to evaluate the pavement surface conditions of the plastic roads. The study is made only on the following distress such as crack, potholes, ravelling and patches, hence the proposed model is limited to predict the mentioned distress alone. Similarly, the model is valid only in the southern Indian road conditions which reflect the bitumen material characteristics supporting the hot climatic conditions. The research can be extended to consider all types of pavement surface distress and roads all over the India with all climatic conditions and various material characteristics.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eA. wrote the manuscriptR.J.Modelling workC.Supervision of WorkR.Figure preparationEsnaashari . Grammer checkFern\u0026aacute;ndez-Campusano. Supervision\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe datasets generated and analyzed during the current study are not publicly available due but are available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eMinistry of Road Transport \u0026amp; Highways. \u003cem\u003eYear End Review 2025 \u0026ndash; Ministry of Road Transport \u0026amp; Highways\u003c/em\u003e. (2025).\u003c/li\u003e\n \u003cli\u003eGuo, R. \u003cem\u003eet al.\u003c/em\u003e Ultra-thin asphalt pavement preservation layer: Performance enhancement mechanism, key technologies, and development trends. \u003cem\u003eCase Studies in Construction Materials\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, (2025).\u003c/li\u003e\n \u003cli\u003eSiva Rama Krishna, U. \u0026amp; Naga Satish Kumar, C. A case study on maintenance of bituminous concrete pavement considering life cycle cost analysis and carbon footprint estimation. \u003cem\u003eInternational Journal of Construction Management\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 1756\u0026ndash;1764 (2022).\u003c/li\u003e\n \u003cli\u003eMinistry of Rural Development. \u003cem\u003eConstruction of Roads Using Plastic Wastes\u003c/em\u003e. (2025).\u003c/li\u003e\n \u003cli\u003ePRS LEGISLATIVE RESEARCH. \u003cem\u003eDemand for Grants 2024-25: Road Transport and Highways\u0026nbsp;\u003c/em\u003e. (2024).\u003c/li\u003e\n \u003cli\u003ePatil Prathisthan\u0026rsquo;, D. Y. \u0026amp; Jalindar, P. Effects of Bad Drainage on Roads Patil Abhijit (Corresponding author). www.iiste.orgwww.iiste.org.\u003c/li\u003e\n \u003cli\u003eNautiyal, A. \u0026amp; Sharma, S. Scientific approach using AHP to prioritize low volume rural roads for pavement maintenance. \u003cem\u003eJ. Qual. Maint. Eng.\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 411\u0026ndash;429 (2022).\u003c/li\u003e\n \u003cli\u003eGertler, P. J., Gonzalez-Navarro, M., Gračner, T. \u0026amp; Rothenberg, A. D. Road maintenance and local economic development: Evidence from Indonesia\u0026rsquo;s highways. \u003cem\u003eJ. Urban Econ.\u003c/em\u003e \u003cstrong\u003e143\u003c/strong\u003e, (2024).\u003c/li\u003e\n \u003cli\u003eVandana Tare, H. S. G. A. B. K. M. ram. PAVEMENT DETERIORATION MODELING FOR LOW VOLUME ROADS. \u003cem\u003eJournal Of The Indian Roads Congress Volume\u003c/em\u003e \u003cstrong\u003e73\u003c/strong\u003e, 67\u0026ndash;891 (2013).\u003c/li\u003e\n \u003cli\u003eRandhawa, S. \u003cem\u003eet al.\u003c/em\u003e Paved or unpaved? A deep learning derived road surface global dataset from mapillary street-view imagery. \u003cem\u003eISPRS Journal of Photogrammetry and Remote Sensing\u003c/em\u003e \u003cstrong\u003e223\u003c/strong\u003e, 362\u0026ndash;374 (2025).\u003c/li\u003e\n \u003cli\u003eAlmnam, A. \u0026amp; Akhmira, A. \u003cem\u003eASSESSMENT OF ROAD PAVEMENT CONDITION USING PAVEMENT CONDITION INDEX (PCI)\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eShtayat, A. \u003cem\u003eet al.\u003c/em\u003e Optimizing Road Pavement Assessment Using Advanced Image Processing Techniques. \u003cem\u003eSustainability (Switzerland)\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, (2025).\u003c/li\u003e\n \u003cli\u003eKhatoon, A. \u003cem\u003eet al.\u003c/em\u003e Advancement in Pavement Condition Assessment: An AI-Based Computer Vision Approach. in \u003cem\u003e2024 Multimedia University Engineering Conference, MECON 2024\u003c/em\u003e (Institute of Electrical and Electronics Engineers Inc., 2024). doi:10.1109/MECON62796.2024.10776158.\u003c/li\u003e\n \u003cli\u003eRadwan, M. M., Faris, S. A., Barakat, A. Y. \u0026amp; Mousa, A. Distress-Based Pavement Condition Assessment Using Artificial Intelligence: A Case Study of Egyptian Roads. \u003cem\u003eEng\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, (2025).\u003c/li\u003e\n \u003cli\u003eAdnan, T. \u0026amp; Erfani, A. Explainable AI for predicting pavement roughness under maintenance and no-maintenance scenarios. \u003cem\u003eResults in Engineering\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, (2026).\u003c/li\u003e\n \u003cli\u003eIbragimov, E., Kim, Y., Lee, J. H., Cho, J. \u0026amp; Lee, J. J. Automated Pavement Condition Index Assessment with Deep Learning and Image Analysis: An End-to-End Approach. \u003cem\u003eSensors\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, (2024).\u003c/li\u003e\n \u003cli\u003eFakhri, M., Pourjafar, S. V. \u0026amp; Daneshvari, M. H. Texture-based image analysis and explainable machine learning for polished asphalt identification in pavement condition monitoring. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, (2025).\u003c/li\u003e\n \u003cli\u003eZuo, D., Shao, L., Chen, B., Tan, J. \u0026amp; Yang, J. Pavement damage detection and evaluation based on UAV image and improved AHP. \u003cem\u003eCase Studies in Construction Materials\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, (2025).\u003c/li\u003e\n \u003cli\u003eWashington DC. USING ARTIFICIAL INTELLIGENCE TO EVALUATE PAVEMENT CONDITION AND SAFETY. \u003cem\u003eFederal Highway Administration\u003c/em\u003e 1 (2024) doi:10.21949/1521492.\u003c/li\u003e\n \u003cli\u003eSaleh Abu Dabous, M. A. G. W. Z. K. H. R. A.-R. Artificial intelligence applications in pavement infrastructure damage detection with automated three-dimensional imaging- A systematic review. \u003cem\u003eAlexandria Engineering Journal\u003c/em\u003e \u003cstrong\u003e117\u003c/strong\u003e, 510\u0026ndash;533 (2025).\u003c/li\u003e\n \u003cli\u003eIRC 82. \u003cem\u003eCODE OF PRACTICE FOR MAINTENANCE OF BITUMINOUS ROAD SURFACES INDIAN ROADS CONGRESS\u003c/em\u003e. https://www.irc.nic.in/ (2015).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Flexible Pavement, Pavement Condition Rating (PCR), AI Model, Distress","lastPublishedDoi":"10.21203/rs.3.rs-9207586/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9207586/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eDeveloping countries were struggling to maintain its road networks in fit surface conditions due to adherence of the conventional method of evaluating the pavement surface condition using multiple physical equipment. The conventional method to measure the multiple distresses is a time consuming and costlier process. Hence in this research, we propose a deep learning-based Pavement Surface Condition Rating of plastic roads. The proposed hybrid model was adopted to assess the pavement surface condition assessment aligned to Pavement condition rating recommended by IRC 82:2015 used in pavement maintenance of Indian roads. In this research, 4k video image of pavement having distresses such as crack, potholes, patches and raveling are acquired by unmanned aerial vehicle mounted with camera. The collected video dataset was used to develop distress prediction model using YOLOv8 integrated with Gray-Level Co-occurrence Matrix (GLCM) and TAMURA feature extraction algorithm. It is observed from the results that the pavement condition rating based on conventional and deep learning- based Image processing technique has high degree of correlation.\u003c/p\u003e","manuscriptTitle":"Pavement Surface Condition Rating of Flexible Pavement based on Artificial Intelligence based Deep Learning Technique","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-14 09:36:08","doi":"10.21203/rs.3.rs-9207586/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"69369d76-50e3-4f11-b06f-303d706310e4","owner":[],"postedDate":"April 14th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":66156643,"name":"Physical sciences/Engineering"},{"id":66156644,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-04-28T13:16:43+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-14 09:36:08","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9207586","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9207586","identity":"rs-9207586","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}