An Automated and Fine- Tuned Image Detection and Classification System for Plant Leaf Diseases

Abstract

Plant leaf disease detection is the process of discovering and diagnosing diseases that damage the plant leaves. This can be done using a variety of approaches including visual inspection, laboratory tests, computer vision techniques etc. Plant leaf diseases must be detected and categorize the cause in order to take corresponding counter measures to manage and control them for healthy and achieve high-yielding crops. In this contemporary Deep Learning era traditional object detection methods have become obsolete due to limitations such as the need for manual crafted features, lack of robustness, and inability to handle large datasets. Deep learning-based approaches are more robust, can account for differences in brightness, interference, and perspective, and can automatically learn features from enormous datasets. These techniques are practical because they can efficiently process large datasets in a brief time frame. Several research studies have developed expert and automatic disease detection methods. Most of this research are limited to one or two plant species and a few disease types. In this research study, we presented a computer vision approach applying the state-of-the-art YOLO algorithm with our own dataset consisting of 8 different classes of pant leaf diseases caused by fungi, bacteria, viruses and pest. Our system achieved promising results and effectively predicted the diseases with the bounding boxes and class probabilities. Overall, this research paper provides a solution for plant disease detection using YOLOv5 and contributes to the development of automated agricultural management systems.