Table 4 The result of counting comparison between improved points REP, RepLKNet; IM, implicitly decoupled detection head; ASL, Asymmetric Loss.
From: A generalized model for accurate wheat spike detection and counting in complex scenarios
| Â | Model | RMSE | R2(%) | MAE | MSE |
|---|---|---|---|---|---|
Ablation experiments | YOLOX | 11.21 | 76.44 | 6.03 | 125.66 |
| Â | REP | 10.96 | 77.47 | 5.54 | 120.12 |
| Â | IM | 12.68 | 69.84 | 6.41 | 160.78 |
| Â | ASL | 11.72 | 74.26 | 6.57 | 137.35 |
|  | REP + ASL | 9.88 | 81.69 | 6.06 | 97.61 |
| Â | RIA-SpikeNet(our) | 7.19 | 90.29 | 5.25 | 51.69 |
| Â | YOLOv4 | 16.86 | 44.60 | 8.93 | 284.25 |
| Â | YOLOv5 | 13.27 | 66.05 | 6.88 | 176.09 |
| Â | YOLOR | 8.21 | 87.14 | 6.14 | 67.40 |
Comparison experiment | GWC_YOLOv5 | 10.90 | 76.99 | 6.76 | 118.81 |
| Â | YOLOv6 | 8.60 | 86.10 | 6.02 | 66.42 |
| Â | YOLOv7 | 11.59 | 74.22 | 6.72 | 134.32 |
| Â | YOLOv8 | 8.15 | 87.53 | 5.53 | 65.85 |
| Â | RIA-SpikeNet(our) | 7.19 | 90.29 | 5.25 | 51.69 |
