Ball Detection on ISSIA-CNR Soccer Dataset (test)
87.7APDeepBall
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DeepBallNo. of trainable parameters=48 658, FPS=1902019.02 | 87.7 | 95.1 | |
| Reno et al.No. of trainable parameters=313 922, FPS=322019.02 | 83.4 | 91.7 | |
| DeepBall (no hypercolumns/context)Hypercolumns=false, Context=false, No. of trainable parameters=29 146, FPS=2702019.02 | 83.3 | 91.1 | |
| DeepBall (no data augmentation)Data Augmentation=false, No. of trainable parameters=48 658, FPS=1902019.02 | 79.2 | 89.9 | |
| Speck et al.No. of trainable parameters=332 365 744, FPS=222019.02 | 22 | 22 |