Table 6 Performance comparison of four object detection networks.
From: Real-time detection of particleboard surface defects based on improved YOLOV5 target detection
Model | Detection of single image time (s) | Model weight size (MB) | Model parameter size | mAP (%) |
|---|---|---|---|---|
Our model (PB-YOLOv5s) | 0.031 | 5.4 | 2.55 × 106 | 94.40 |
YOLOv4 | 0.069 | 235.0 | 6.39 × 107 | 89.71 |
YOLOv3 | 0.087 | 244.0 | 6.27 × 107 | 86.42 |
Faster RCNN | 0.256 | 512.0 | 2.40 × 106 | 81.35 |
