Comparison of Pre-trained Deep Model Using Tomato Leaf Disease Classification System

Our country has vast potential to emerge as a cardinal exporter of agricultural products, but lack of rapid quality assessment techniques, enormous losses in processing and handling after harvesting, crop disease, etc. which results in a lower contribution to the global market. Diseases harm the health of the crop, which in turn affects the growth of the plant. It is important to track its growth in order to ensure minimal losses to the cultivated crop. Nowadays, automation in agriculture is very much needed, and a lot of work is being carried on the same. To get the early benefits to farmers is to diagnose the problem early with the help of fast, reliable, non-destructive method. Images of tomato leaves (three diseases and a healthy class) are used from the Plant Village dataset which is provided as input to three architectures based on deep learning, namely Inception-v3, ResNet-50, and Inception-ResNet-v2.