Tomato Leaf Disease Identification Based on Yolov8

To quickly identify tomato leaf diseases in the planting environment, based on the YOLOv8 model, this study explores the improvement of four types of loss functions compared to the original YOLOv8 model, and replaces the backbone network of YOLOv8 to further improve the accuracy of model recognition. The improved PIoU v2 loss function has increased the Precision value by 1.1 percentage points, the Recall value by 2.8 percentage points, the mAP value by 1.3 percentage points, and the FasterNet backbone has improved the detection performance of the model, with a 0.4 percentage point increase in Precision value and a 0.3 percentage point increase in Recall value compared to the CIOU loss function used in the original model. The mAP value increased by 0.2 percentage points. Compared to the original model, the improvement effect has been improved.