Method Details
Details for method 'ResNet-38'
Method overview
| name | ResNet-38 |
| challenge | pixel-level semantic labeling |
| details | single model, single scale, no post-processing with CRFs Model A2, 2 conv., fine only, single scale testing The submissions was previously listed as "Model A2, 2 conv.". The name was changed for consistency with the other submission of the same work. |
| publication | Wider or Deeper: Revisiting the ResNet Model for Visual Recognition Zifeng Wu, Chunhua Shen, Anton van den Hengel arxiv https://arxiv.org/abs/1611.10080 |
| project page / code | https://github.com/itijyou/ademxapp |
| used Cityscapes data | fine annotations |
| used external data | ImageNet, Places |
| runtime | n/a |
| subsampling | no |
| submission date | November, 2016 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 78.3673 |
| iIoU Classes | 59.1052 |
| IoU Categories | 90.9364 |
| iIoU Categories | 81.0773 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.5183 | - |
| sidewalk | 85.7477 | - |
| building | 93.0519 | - |
| wall | 55.4868 | - |
| fence | 59.1284 | - |
| pole | 67.0935 | - |
| traffic light | 74.7631 | - |
| traffic sign | 78.696 | - |
| vegetation | 93.7342 | - |
| terrain | 72.5751 | - |
| sky | 95.5241 | - |
| person | 86.5724 | 71.8885 |
| rider | 69.2288 | 50.6193 |
| car | 95.7141 | 90.5325 |
| truck | 64.5401 | 41.9782 |
| bus | 78.7866 | 54.1653 |
| train | 74.0624 | 51.3811 |
| motorcycle | 69.034 | 48.1091 |
| bicycle | 76.7206 | 64.168 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.7168 | - |
| nature | 93.3772 | - |
| object | 73.3708 | - |
| sky | 95.5241 | - |
| construction | 93.4566 | - |
| human | 87.0009 | 73.1989 |
| vehicle | 95.1083 | 88.9557 |