Table 5. Result for frame-based one-class polyp detection.
Precision | Recall | F1 score | AP | Inference Time | FPS | |
---|---|---|---|---|---|---|
Faster RCNN | 63.9 | 89.8 | 74.7 | 85.6 | 52ms | 19 |
SSD | 91.3 | 82.0 | 86.4 | 86.3 | 17ms | 59 |
YOLOv3 | 95.9 | 78.0 | 86.0 | 81.0 | 17ms | 59 |
RetinaNet | 86.1 | 86.6 | 86.3 | 87.9 | 61ms | 16 |
DetNet | 85.8 | 81.8 | 83.7 | 80.5 | 64ms | 16 |
RefineDet | 91.2 | 86.2 | 88.6 | 88.5 | 31ms | 32 |
YOLOv4 | 89.8 | 74.4 | 81.3 | 83.9 | 30ms | 33 |
ATSS | 92.1 | 84.7 | 88.3 | 88.1 | 53ms | 19 |