TABLE 3.
Performance comparison of detection and segmentation methods.
| Methods | Appledataset | Persimmondataset | Average time/s | ||
|
|
|
||||
| mAP | mAPS | mAP | mAP S | ||
| SOLO | — | 57.4 | — | 76.5 | 0.49 |
| PolarMask | 56.3 | 53.8 | 68.3 | 66.1 | 0.52 |
| YOLACT | 57.6 | 60.5 | 66.9 | 71.2 | 0.45 |
| TensorMask | — | 65.3 | — | 72.4 | 0.71 |
| FCOS | 61.2 | — | 78.6 | — | 0.42 |
| RetinaMask | 69.2 | 69.4 | 77.5 | 73.1 | 0.92 |
| MaskR-CNN | 70.3 | 70.9 | 79.9 | 79.3 | 0.48 |
| MSR-CNN | 71.2 | 71.6 | 80.3 | 80.9 | 0.56 |
| OURS | 71.0 | 71.4 | 81.8 | 82.6 | 0.23 |