Improving Ground Cover Crop Fractional Vegetation Mapping via Causality-Based Deep Representation Learning

Semantic segmentation and deep learning methods have rarely been applied to Fractional Vegetation Cover (FVC) segmentation tasks due to the lack of publicly available datasets for training deep learning models. FVC is a key indicator for assessing vegetation distribution, crop density, and crop responses to water availability and fertilizer application, yet conventional field-based measurement methods are time-consuming, costly, labor intensive, and may lack the accuracy required for critical applications such as drought stress evaluation, and water productivity. In this paper, we introduced causality-based deep learning technique for FVC segmentation on a publicly available RGB dataset that consists of four ground cover crops, Phyla nodiflora L, Cynodon dactylon, Frankenia thymifolia Desf, and Oxalis stricta L. By separating causal from spurious correlations in pretrained features, the stepwise intervention and reweighting (SIR) method reduced confounding bias and enhanced generalization across tasks and datasets. Extensive experiments on the FVC dataset, both with and without causality learning, showed that the proposed U-Net + VGG16 model with causality learning achieved an accuracy of 92.04%, precision of 92.31%, recall of 92.65%, and an F1-score of 91.98%, outperforming non-causal baselines.