Anthracnose Detection and Classification in Chilli Leaves Using Mobile Net: UAV Images

The health of a plant has remained an indispensable factor for sustainable and enriching crop production to fulfill the Nation's food demands. A degradation in the health of the crop leads to low crop yields. Out of a wide range of factors affecting the health of the crop the most concerning factor remains the plant diseases and pests. More than 50% of India's population is engaged in agricultural practices. In the agricultural sector, India is among the top 3 exporters of Dried Chilli. However, Chilli exports have faced a descent in recent years owing to crop yield loss due to diseases caused by fungi or pests. Anthracnose is the most prevalent disease in Chilli plants. The process of anthracnose disease detection in Chilli crops is a tedious process due to a lack of laboratory facilities and expertise. Such a situation demands for development of an 'automated detection of anthracnose disease in Chilli plant' system which puts the process of early and precise detection of plant disease at ease. Computer Vision advancements make it possible for automatic disease detection using visible symptoms on the crop. Keeping this in view, this paper proposes an automated model for the detection of anthracnose in the Chilli plant using a Mobile Net-based convolutional neural network (CNN). Over the last few years researchers such as Boulent et al. stated in their article that the integration of aerial image data and AI-based methods holds promise for offering an effective, high-throughput technique for detecting plant diseases in real-world conditions. Hence the dataset utilized for our model is aerial images using UAV (unnamed aerial vehicle). The proposed system for anthracnose detection in Chilli developed using Mobile Net achieved an accuracy of 99.6%.