Rice Crop Disease Detection Using Explainable AI

The development of automated systems that use Artificial Intelligence and Deep Learning Techniques has enabled the identification and detection of diseases in images with increasing accuracy and efficiency. In today’s scenario, farmers are facing many problems. They face challenges like poor infrastructure, lack of knowledge of emerging technologies, poor soil quality, investment in farm production, etc., when growing rice crops. In South India, many farmers need to know how to select fertilizers and pesticides for their rice crops properly. Due to the diseases that affect the rice crop, farmers and consumers face a significant loss. Nearly 60% of the farmers face losses in crop yield. As a result, there have been multiple reports of the deaths of the farmers. We identified the rice crop diseases and focused on specific aspects, including image dataset, no. of diseases (classes), and model accuracy. We created a model using Convolution Neural Network (CNN) for classifying images and explainable Artificial Intelligence (xAI) by using a local interpretability technique called LIME (Local Interpretable Model-Agnostic Explanations) to explain the predictions that the model makes. This model can extract the features of the disease in the given image. Evaluation of the images from the Rice Disease Image dataset shows that our model’s accuracy is 91.60%.