Advancing Black Gram (Vigna Mungo L.) Disease Diagnosis Through Deep Learning Techniques
Interestingly, in sustainable crop protection, disease diagnosis, and management are crucial in sustainable crop production. It plays a captious role in rain-fed pulses because the occurrence of season of cropping, cultivation after main crop, availability of soil moisture in poor conditions, consecutively following the same cultivars are acting a predominant role in disease diagnosis approaches and confirmation. Under these situations, occurrence of the manual errors (or) mis find faults resulting in complete drawbacks to disease diagnosis and management for farmers and scientists worldwide. Keeping this background, applying deep learning techniques is most helpful in diagnosing plant diseases silently and superiorly. Deep learning techniques were carried out in this study to diagnose foliar diseases in black gram such as anthracnose, leaf crinkle, powdery mildew, and yellow mosaic that causes a severe yield loss (>50%) silently accompanied by green biomass. A vast field survey was conducted in the black gram growing Cauvery delta zone of four blocks in Pudukkottai district, Tamil Nadu, India, with 27376 images collected. Furthermore, the advanced inception V3 model has been used for analysis, assessment, and prediction for the diagnosis of diseases. The model was investigated with 20 percent, 40 percent, and 50 percent dropout rates. The result showed that an Inception V3 model, with a 20 percent dropout rate, gave the best performance with an accuracy of 99.22 percent and a loss of 0.0249. The high performance rate shows automated disease diagnosis, which helps the farmers develop disease management strategies at the preliminary stages of their growth.
