Method Details
Details for method 'SSAP'
Method overview
| name | SSAP |
| challenge | panoptic semantic labeling |
| details | SSAP, ResNet-101, Cityscapes fine-only |
| publication | SSAP: Single-Shot Instance Segmentation With Affinity Pyramid Naiyu Gao, Yanhu Shan, Yupei Wang, Xin Zhao, Yinan Yu, Ming Yang, Kaiqi Huang ICCV 2019 https://arxiv.org/abs/1909.01616 |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | n/a |
| subsampling | no |
| submission date | September, 2019 |
| previous submissions |
Average results
| Metric | All | Things | Stuff |
|---|---|---|---|
| PQ | 58.8917 | 48.4133 | 66.5123 |
| SQ | 82.4067 | 82.9244 | 82.0301 |
| RQ | 70.6124 | 58.2701 | 79.5885 |
Class results
| Class | PQ | SQ | RQ |
|---|---|---|---|
| road | 97.9862 | 98.1155 | 99.8682 |
| sidewalk | 75.8626 | 84.7733 | 89.4888 |
| building | 88.0594 | 90.7205 | 97.0667 |
| wall | 34.7153 | 74.8441 | 46.3835 |
| fence | 38.0181 | 74.6416 | 50.9341 |
| pole | 56.1156 | 68.021 | 82.4975 |
| traffic light | 51.562 | 72.3644 | 71.2533 |
| traffic sign | 68.7725 | 76.8301 | 89.5125 |
| vegetation | 89.6503 | 91.41 | 98.075 |
| terrain | 43.5707 | 78.7707 | 55.3134 |
| sky | 87.3221 | 91.8399 | 95.0808 |
| person | 50.2568 | 81.3019 | 61.815 |
| rider | 46.1369 | 76.3007 | 60.4672 |
| car | 65.5824 | 88.2224 | 74.3376 |
| truck | 43.2086 | 91.4876 | 47.2289 |
| bus | 54.8392 | 91.6935 | 59.8071 |
| train | 48.0974 | 83.8908 | 57.3333 |
| motorcycle | 41.919 | 76.1498 | 55.0481 |
| bicycle | 37.2665 | 74.3488 | 50.1238 |