Method Details
Details for method 'Deep Affinity Net [fine-only]'
Method overview
| name | Deep Affinity Net [fine-only] |
| challenge | instance-level semantic labeling |
| details | A proposal-free method that uses FPN generated features and network predicted 4-neighbor affinities to reconstruct instance segments. During inference time, an efficient graph partitioning algorithm, Cascade-GAEC, is introduced to overcome the long execution time in the high-resolution graph partitioning problem. |
| publication | Deep Affinity Net: Instance Segmentation via Affinity Xingqian Xu, Mangtik Chiu, Thomas Huang, Honghui Shi https://arxiv.org/abs/2003.06849 |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | n/a |
| subsampling | no |
| submission date | March, 2020 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| AP | 27.5109 |
| AP50% | 48.0132 |
| AP100m | 41.5132 |
| AP50m | 46.8644 |
Class results
| Class | AP | AP50% | AP100m | AP50m |
|---|---|---|---|---|
| person | 24.5089 | 51.4238 | 39.4186 | 39.5929 |
| rider | 22.1552 | 53.1593 | 35.1234 | 35.9763 |
| car | 43.654 | 66.7329 | 62.9683 | 66.4922 |
| truck | 29.4808 | 38.794 | 43.439 | 54.6055 |
| bus | 38.2744 | 51.1644 | 61.2321 | 76.5504 |
| train | 31.9255 | 49.8165 | 46.2889 | 56.7742 |
| motorcycle | 18.0044 | 40.2176 | 24.8552 | 26.1501 |
| bicycle | 12.0839 | 32.7968 | 18.7797 | 18.7735 |