Water Stress Identification in Chickpea Images using Machine Learning

Computer vision based plant phenotyping methods can be employed for plant health monitoring and plant disease detection. This can play a crucial role in precision farming, enabling increased crop outputs to meet the demands of a rising world population. In this work, we consider an important problem of quantifying the moisture stress levels in chickpea plants using the shoot images. Compared to leave images, using shoot images is non-destructive and provides more information. Pulses like chickpea, owing to their high protein and nutrition content, can play an important role towards ensuring food security in poor countries. In the present work, we have employed Machine Learning techniques to compare the performance of different descriptors for stress classification induced by water deficiency from chickpea shoot images. From the results, we found that the support vector machine classifier outperforms other classifiers using SIFT descriptors, achieving nearly 73% classification accuracy.