Method Details
Details for method 'LGE A&B Center: HANet (ResNext-101)'
Method overview
| name | LGE A&B Center: HANet (ResNext-101) |
| challenge | pixel-level semantic labeling |
| details | Dataset: "fine train + fine val + coarse", Backbone: Mapillary pretrained ResNext-101 |
| publication | Cars Can’t Fly up in the Sky: Improving Urban-Scene Segmentation via Height-driven Attention Networks Sungha Choi (LGE, Korea Univ.), Joanne T. Kim (Korea Univ.), Jaegul Choo (KAIST) CVPR 2020 https://arxiv.org/abs/2003.05128 |
| project page / code | https://github.com/shachoi/HANet |
| used Cityscapes data | fine annotations, coarse annotations |
| used external data | ImageNet, Mapillary |
| runtime | n/a |
| subsampling | no |
| submission date | March, 2020 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 83.1703 |
| iIoU Classes | 62.625 |
| IoU Categories | 92.067 |
| iIoU Categories | 80.662 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.8312 | - |
| sidewalk | 87.9859 | - |
| building | 94.1956 | - |
| wall | 66.5642 | - |
| fence | 64.814 | - |
| pole | 72.0216 | - |
| traffic light | 78.1983 | - |
| traffic sign | 81.4281 | - |
| vegetation | 94.1574 | - |
| terrain | 74.4767 | - |
| sky | 96.0846 | - |
| person | 88.0757 | 70.6074 |
| rider | 75.6363 | 55.1073 |
| car | 96.5138 | 91.7535 |
| truck | 80.3446 | 47.7202 |
| bus | 93.1892 | 61.1293 |
| train | 86.5642 | 56.1707 |
| motorcycle | 72.4761 | 53.1504 |
| bicycle | 78.678 | 65.3613 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.8044 | - |
| nature | 93.8363 | - |
| object | 77.3083 | - |
| sky | 96.0846 | - |
| construction | 94.3179 | - |
| human | 87.9974 | 71.4354 |
| vehicle | 96.1202 | 89.8887 |