Table 4 The detection accuracy of YOLO models with image augmentation on the validation set.
YOLO Model | Augmentation | Pink (mAP) | Purple (mAP) | Orange (mAP) | Average (mAP) |
---|---|---|---|---|---|
YOLOv4 | Mosaic + Mix-up | 0.599 | 0.327 | 0.418 | 0.448 |
E&A | 0.457 | 0.177 | 0.514 | 0.383 | |
IEP augmentation | 0.662 | 0.657 | 0.55 | 0.623 | |
YOLOv7 | Mosaic + Mix-up | 0.847 | 0.679 | 0.613 | 0.713 |
E&A | 0.806 | 0.716 | 0.658 | 0.727 | |
IEP augmentation | 0.837 | 0.67 | 0.622 | 0.71 | |
YOLOv8 | Mosaic + Mix-up | 0.691 | 0.837 | 0.576 | 0.701 |
E&A | 0.728 | 0.8 | 0.645 | 0.725 | |
IEP augmentation | 0.899 | 0.674 | 0.612 | 0.728 |