Method Details
Details for method 'R-CNN + MCG convex hull'
Method overview
| name | R-CNN + MCG convex hull |
| challenge | instance-level semantic labeling |
| details | We compute MCG object proposals [1] and use their convex hulls as instance candidates. These proposals are scored by a Fast R-CNN detector [2]. [1] P. Arbelaez, J. Pont-Tuset, J. Barron, F. Marqués, and J. Malik. Multiscale combinatorial grouping. In CVPR, 2014. [2] R. Girshick. Fast R-CNN. In ICCV, 2015. |
| publication | The Cityscapes Dataset for Semantic Urban Scene Understanding M. Cordts, M. Omran, S. Ramos, T. Rehfeld, M. Enzweiler, R. Benenson, U. Franke, S. Roth, B. Schiele CVPR 2016 https://www.cityscapes-dataset.com/wordpress/wp-content/papercite-data/pdf/cordts2016cityscapes.pdf |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | 60 s |
| subsampling | 2 |
| submission date | April, 2016 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| AP | 4.55117 |
| AP50% | 12.9045 |
| AP100m | 7.72122 |
| AP50m | 10.2591 |
Class results
| Class | AP | AP50% | AP100m | AP50m |
|---|---|---|---|---|
| person | 1.31034 | 5.60813 | 2.57423 | 2.7342 |
| rider | 0.60637 | 3.91998 | 1.06862 | 1.08409 |
| car | 10.5089 | 26.0016 | 17.4514 | 21.1824 |
| truck | 6.14297 | 13.8018 | 10.6379 | 13.9857 |
| bus | 9.69777 | 26.3358 | 17.353 | 25.2147 |
| train | 5.85429 | 15.8461 | 9.15934 | 14.1514 |
| motorcycle | 1.7459 | 8.63505 | 2.59719 | 2.73806 |
| bicycle | 0.542845 | 3.0876 | 0.92805 | 0.982185 |