State-of-Art Deep Learning Based Tomato Leaf Disease Detection

Abstract

In India, the tomato plant is a popular staple food with high commercial value and considerable production capacity; however, the quality and quantity of the tomato harvest decreases due to a variety of diseases and henceforth leads to great financial loss for farmers. With lack of agricultural professions to assist the farmers, a deep learning (DL) based user friendly, just-in-time mobile is proposed for the detection of crop diseases for assisting farmers to know about the type of tomato disease and the remedy for the same. Two DL based methods: YOLO and Faster RNN have been used for detection; followed by classification using SVM and Random forest tree. YOLO and Random forest tree resulted in accuracy in the range of 90% to 95%. The developed app provides option to the farmer to operate in English as well as in local language Kannada of Karnataka state of India.