Bounding box regression
| Model | IoU | Accuracy | Time_sec | Loss | Image_size | Bce_coef | Augmentation | Batch_size | Learning_rate |
|---|---|---|---|---|---|---|---|---|---|
| mobilenetv2_120d | 0.8906 | 0.9947 | 0.33 | EfficientIoU | 512 | 0.1 | medium | 8 | 3e-4 |
| mobilenetv2_120d | 0.8669 | 0.9920 | 0.083 | EfficientIoU | 224 | 0.1 | medium | 8 | 3e-4 |
| mobilenetv2_100 | 0.8403 | 0.9947 | 0.063 | EfficientIoU | 224 | 0.1 | medium | 8 | 3e-4 |
| mobilenetv2_110d | 0.8524 | 0.9947 | 0.066 | EfficientIoU | 224 | 0.1 | medium | 8 | 3e-4 |
| mobilenetv2_140 | 0.8498 | 0.8989 | 0.069 | EfficientIoU | 224 | 0.1 | medium | 8 | 3e-4 |
| mobilenetv2_110d | 0.8543 | 0.9947 | 0.066 | EfficientIoU | 224 | 0.1 | hard | 8 | 3e-4 |
| mobilenetv2_110d | 0.8570 | 0.9920 | 0.066 | EfficientIoU | 224 | 0.05 | hard | 8 | 3e-4 |
| mobilenetv2_110d | 0.8593 | 0.9947 | 0.066 | EfficientIoU | 224 | 0.15 | hard | 8 | 3e-4 |