An Automated Plant Disease Detection Method Using Image Processing and Machine Learning
Bilal Ahmed Sheikh, Department of Computer Sciences, University of Engineering and Technology, Taxila, Pakistan.
Abid Rauf, Department of Computer Sciences, University of Engineering and Technology, Taxila, Pakistan.
Marriam Nawaz, Department of Software Engineering, University of Engineering and Technology, Taxila, Pakistan.
Corresponding Author:
Bilal Ahmed Sheikh (sheikh.bilal01@gmail.com)
Abstract:
The development of agriculture is crucial for a nation's economy. However, plant diseases pose a significant challenge to the quantity and quality of crop growth. Hence, early detection of diseases is crucial to protect the crops. Traditionally, the farmers perform optical inspection of crops, which not only consumes a lot of time but can also result in misidentification due to similarity of diseases. Hence, there is a need for automated systems that can precisely detect diseases from leaves which will save time and produce better results than the traditional approaches. Traditional methods have failed to achieve robust performance due to the use of poor-quality databases and lack of preprocessing. In this study, we propose an automated plant disease detection method using Machine Learning (ML) and Computer Vision (CV) approach. Initially, we preprocess the images to remove noise from the images. The images are then segmented by removing the background pixels and keeping only leaf portion. Then various features such as Local Binary Pattern (LBP), Histogram of Gradients (HOG) and Maximally Stable Extremal Regions (MSER) are extracted from these images. These features are then classified via Support Vector Machines (SVM), K-Nearest Neighbor (KNN) and Decision Trees (DT). The suggested study resulted in maximum accuracy of 99.3% via SVM. The suggested framework is trained and validated on a publicly available Plant Village database containing leaf images from a variety of crops. The results prove efficacy and robustness of our proposed method; hence, it can be utilized in real-time environment to accurately detect the presence of plant diseases from leaf images.
Keywords:
Plant Disease Detection; Machine Learning; Computer Vision; Plant Village