Automated Plant Disease Detection Using Deep Learning
Dr. Dwiti Krishna Bebarta1, Gajjarapu Bhavya2, Tumu Swapna3, Chadalavada Sri Malvina4, Gangiredla Asha Deepthi
Naidu5
1Associate Professor, Department of Computer Science and Engineering, Gayatri Vidya Parishad college of Engineering for Women, Visakhapatnam, Andhra Pradesh, India.
2,3,4,5Research Scholar, department of Computer Science and Engineering, Gayatri Vidya Parishad College of Engineering for Women, Visakhapatnam, Andhra Pradesh, India.
dkbebarta@gvpcew.ac.in, bhavyagajjarapu143@gmail.com, swapna.tumu2000@gmail.com, srimalvinachadalavada@gmail.com, ashagangiredla@gmail.com
Abstract— Traditional methods of diagnosing plant diseases are mainly based on expert diagnosis which easily causes delay in crop disease control and crop management. Due to the problems of many target areas and similar target types in the process of plant disease detection, the identification accuracy and speed are required to be high. Therefore, it is necessary to optimize and improve the existing methods (CNN, RCNN, Fast RCNN, Faster RCNN, and SSD) to meet the detection needs. So, we came up with a deep learning-based approach to identify the plant leaf diseases and classify the diseases using an object detection model called YOLO. There are different versions of YOLO, proposed in recent times, in that YOLOv5 model is considered one of the best models and the other is the recent version, YOLOv8 which was proposed in 2022. So, in this paper we compared the two models of YOLO, YOLOv5 and YOLOv8 on the same dataset, and found that for the dataset used the YOLOv5 model was found to be the better model with a mAP of 0.641, while the YOLOv8 model has mAP of 0.516. This study proposes that the YOLOv5 model is suitable for plant disease identification tasks by comparing it with the latest version of the YOLOv8 model which was proposed in 2022. To make the YOLO model to be better it can be optimized and the transfer learning ability of the model can be used to expand the application scope in the future.
Keywords— Deep Learning, CNN, RCNN, Fast RCNN, Faster RCNN, SSD, YOLO, mAP, Transfer Learning, Disease Classification, Detection.
