期刊名称：International Journal of Early Childhood Special Education
电子版ISSN：1308-5581
出版年度：2022
卷号：14
期号：5
页码：1546-1553
DOI：10.9756/INTJECSE/V14I5.156
语种：English
出版社：International Journal of Early Childhood Special Education
摘要：Unpredictable climatic conditions can greatly affect the crops in terms of vulnerability towards various infections which may be fungal, bacterial, etc. Detection of plant diseases during the initial stages of its spread is crucial to stop its spread which can infect the healthy ones as well. Another major benefit of detecting the disease early is to safeguard farmers from huge losses. It is very difficult to identify leaf diseases with the naked eye. To overcome this problem farmers all over the world spray pesticides. This leads to the spoilage of plants if used in excess. Thus, a proper identification mechanism needs to be established to predict the diseases at an early stage. The existing approaches to this problem use various versions of Regional Convolutional Neural Network (R-CNN) for leaf disease detection. It is a two-stage detection algorithm which makes it expensive in terms of time. The proposed system uses You Only Look Once (YOLO) object detection algorithm for detecting the diseases. The YOLO algorithm is much faster when compared to other object detection algorithms like RCNN. It requires only one propagation through the neural network to detect objects and provides predictions with accuracy on par with comparatively slower algorithms mentioned earlier. This makes it capable of real-time prediction. Although the dataset used to train the YOLO model consists of images of healthy and diseased leaves of tomato plants, the proposed model can be trained for leaves of varieties of other plants for disease detection.
