Application of EfficientNetV2 and YoloV5 for tomato leaf disease identification

Abstract

In this study, we selected the 10 most common tomato leaf diseases with reference to the common disease symptoms of tomato, and built a tomato leaf dataset with labelimg After preprocessing the tomato leaves with labeling, we applied two learning models, EfficientNetV2 and YoloV5„ EfficientNetV2 is a convolutional network model just released in April 2021, this model can better classify tomato leaves to an effective degree, and it has a better training speed and efficiency compared to the previous series of models. After image classification with EfficientNetV2, target detection is needed, and since the algorithm for recognition needs to be built on the inspection robot, the target detection algorithm needs to be extremely accurate. This time on the leaf disease recognition used YoloV5 algorithm detection in the target detection training, Yolo model recognition of high accuracy, in the detection of the leaf, training leaf data set is also the best effect, and V5 model deepening widening network, so that AP accuracy is also increasing, and thus in the speed of consumption above is rising, so In the case of equipped inspection robot, the effect is very objective.