Plant Leaf Disease Classification using Convolutional Neural Network on FPGA

Abstract

Leaf damage due to disease is a major source of concern for farmers. India's agriculture-based economy accounts for a considerable share of the country's GDP. This research proposes the utilization of a machine learning based model on different leaf images to detect the disease of four different types of plants which includes Apple, Maize, Potato and Tomato. A Convolutional Neural Network (CNN) model is used to categorize the kind of disease affecting the leaf, which is trained by using the Plant Village dataset. This trained model is deployed on AMD Xilinx's Kria KV260 FPGA using Vitis AI for the identification of the disease. The accuracy of disease classification achieved is around 98.04%.