基于改进YOLOv5s的烟梗物料目标检测算法
|
|
吕佳铭,张峰,罗亚波
|
Improved YOLOv5s based target detection algorithm for tobacco stem material
|
|
Jiaming LV,Feng ZHANG,Yabo LUO
|
|
| 表 1 模块消融实验结果分析 |
| Tab.1 Analysis of results of module ablation experiments |
|
| 模型 | P/% | R/% | mAP@0.50/% | mAP@0.50∶0.95/% | M/MB | GFLOPs | FPS/帧 | | ①YOLOv5s | 77.8 | 93.3 | 90.3 | 89.0 | 13.8 | 16.6 | 212.77 | | ②YOLOv5s+RepViT-m1 | 81.9 | 85.3 | 89.8 | 86.2 | 26.2 | 24 | 144.93 | | ③YOLOv5s+重参数化的RepViT-m1 | 80.3 | 90.4 | 91.5 | 86.5 | 11.4 | 19.9 | 192.31 | | ④YOLOv5s+Dynamic Head | 79.7 | 93.6 | 92.1 | 90.6 | 13.7 | 17.8 | 185.19 | | ⑤YOLOv5s+RepViT-m1+Dynamic Head | 84.1 | 96.4 | 95.8 | 94.3 | 14.2 | 21.8 | 133.33 | | ⑥本研究算法 | 86.2 | 94.1 | 96.1 | 94.7 | 12.1 | 21.3 | 178.57 |
|
|
|