Table 2.
Comparison of our method and other state-of-the-art methods in terms of mAP values (%) on the Pascal VOC 2012 dataset. The best accuracy is shown in boldface.
| Number | Category | Faster R-CNN (%) | SSD (%) | Mask R-CNN (%) | SSOD (ours) (%) |
|---|---|---|---|---|---|
| 1 | aeroplane | 84.9 | 85.6 | 86.5 | 84.3 |
| 2 | bicycle | 79.8 | 80.1 | 83.9 | 87.8 |
| 3 | bird | 74.3 | 70.5 | 77.6 | 79.4 |
| 4 | boat | 53.9 | 57.6 | 58.8 | 73.2 |
| 5 | bottle | 49.8 | 46.2 | 63.7 | 54.5 |
| 6 | bus | 77.5 | 79.4 | 80.4 | 84.3 |
| 7 | car | 75.9 | 76.1 | 78.2 | 86.1 |
| 8 | cat | 88.5 | 89.2 | 87.6 | 87.1 |
| 9 | chair | 45.6 | 53.0 | 56.0 | 60.8 |
| 10 | cow | 77.1 | 77.0 | 80.1 | 85.0 |
| 11 | dining table | 55.3 | 60.8 | 63.2 | 80.2 |
| 12 | dog | 86.9 | 87.0 | 88.0 | 86.6 |
| 13 | horse | 81.7 | 83.1 | 84.4 | 88.1 |
| 14 | motorbike | 80.9 | 82.3 | 85.7 | 84.5 |
| 15 | person | 79.6 | 79.4 | 79.9 | 79.8 |
| 16 | potted plant | 40.1 | 45.9 | 46.3 | 56.2 |
| 17 | sheep | 72.6 | 75.9 | 74.4 | 81.1 |
| 18 | sofa | 60.9 | 69.5 | 66.8 | 80.8 |
| 19 | train | 81.2 | 81.9 | 81.3 | 87.5 |
| 20 | tv/monitor | 61.5 | 67.5 | 75.0 | 80.6 |
| - | mAP | 70.4 | 72.4 | 74.9 | 79.4 |