Table 1 Comparison of lightweight models
From: Lightweight highland barley detection based on improved YOLOv5
Model | P/% | R/% | F1 | \(\hbox {mAP}_{0.5}/\%\) | \(\hbox {Parameter}/10^6\) | FLOPs /G |
|---|---|---|---|---|---|---|
YOLOv5n | 89.1 | 83.1 | 86.0 | 90.3 | 1.76 | 4.1 |
YOLOv5n+Mobilenetv2 | 66.3 | 60.0 | 63.0 | 63.6 | 0.21 | 0.8 |
YOLOv5n+Mobilenetv3_large | 75.7 | 74.9 | 75.3 | 78.3 | 0.98 | 2.5 |
YOLOv5n+Mobilenetv3_small | 68.9 | 70.5 | 69.7 | 72.5 | 0.45 | 0.9 |
YOLOv5n+Shufflenetv2 | 62.7 | 56.6 | 59.5 | 59.0 | 0.22 | 0.5 |
YOLOv5n+DSConv | 90.7 | 85.0 | 87.8 | 92.1 | 1.51 | 3.6 |
YOLOv5n+GhostConv | 92.7 | 86.3 | 89.4 | 93.2 | 1.46 | 3.6 |