PRECISION AGRICULTURE USING COMPUTER VISION

Abstract :

Computer Vision, which relies on Machine Learning, targets different application domains to solve critical real-life problems. The solutions to these problems are often based on the algorithms, which seek to understand and automate tasks that human system can do. The certainty of algorithms to be used is always a huge challenge and so is finding the right Computer Vision resources. A Computer Vision model is a processing block that takes uploaded inputs, like images or videos, and predicts and returns pre-learned concepts or labels. One of the main applications of Computer Vision is in increasing the agricultural yield. The main factor that hinders better yield is the diseases in plants, particularly in leaves, which is the main problem for food scarcity, and detection of these diseases remains difficult due to the lack of the necessary infrastructure. Accurate techniques of computer vision and deep learning in the field of plant leaf disease detection and classification will be a boon for farmers. These changing technologies can be used by farmers for better yield. In Computer vision, the images of plant leaves are used to detect diseases and pests in the plants, just as diagnosed by experts. To initiates this research, a literature review of plant disease/ pest detection and classification was carried out. 28 related research papers were reviewed. Different methods were employed in the classification of plant leaves disease and pest detection, which resulted in good accuracies. The reviewed works have used either ML techniques or DL techniques or both to classify the classes. Some of the work used a single crop and others used more than one crop for classification. In most of the works, the methods were compared with other state-of-the-art methodologies and gave recommendable accuracies. All works have used some pre-processing techniques, and the healthy and diseased plant leaves were classified differently. The review paper written has been communicated to the research journal, INFOCOMP Journal of Computer Science and is under review. Based on this literature review, the banana leaves diseases detection and classification work has been started. The banana leaf dataset was created with healthy and diseased leaves from various locations in Kazhakuttam through the intervention of the Agricultural Office, Kazhakuttam. The dataset was labelled with diseases and healthy labels and trained using KNN and SVM machine learning algorithms. Some pre-processing techniques were used to get better accuracy. The CNN architectures such as AlexNet and ResNet were also used to train and test the dataset of Banana leaves and shows better results. The fields of agricultural automation are monitoring of crop growth, disease prevention, automatic harvesting, quality testing, automated management of modern farms and the monitoring of farmland information with Unmanned Aerial Vehicle (UAV). UAV remote sensing has contributed to an increase in agricultural production and UAVs are generally acknowledged to be advantageous when applied to crop monitoring, protection, management and other farm operations. The existing computer vision technologies can address the deficiencies of traditional monitoring and reduce the difficulty of traditional growth monitoring in terms of time, continuity and cost. Computer vision technology has the advantages of low cost, small error, high efficiency and good robustness and can be dynamically and continuously analyzed. However, the related methods still have limitations, achieving versatility and stability in various complicated situations is still challenging and extensive work will be required in the future in this regard.