Method Details
Details for method 'FRRN'
Method overview
| name | FRRN |
| challenge | pixel-level semantic labeling |
| details | Full-Resolution Residual Networks (FRRN) combine multi-scale context with pixel-level accuracy by using two processing streams within one network: One stream carries information at the full image resolution, enabling precise adherence to segment boundaries. The other stream undergoes a sequence of pooling operations to obtain robust features for recognition. |
| publication | Full-Resolution Residual Networks for Semantic Segmentation in Street Scenes Tobias Pohlen, Alexander Hermans, Markus Mathias, Bastian Leibe Arxiv https://arxiv.org/abs/1611.08323 |
| project page / code | https://github.com/TobyPDE/FRRN |
| used Cityscapes data | fine annotations |
| used external data | |
| runtime | n/a |
| subsampling | 2 |
| submission date | November, 2016 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 71.8384 |
| iIoU Classes | 45.4616 |
| IoU Categories | 88.8987 |
| iIoU Categories | 75.1365 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.1738 | - |
| sidewalk | 83.3382 | - |
| building | 91.6157 | - |
| wall | 45.8264 | - |
| fence | 51.0851 | - |
| pole | 62.2036 | - |
| traffic light | 69.3919 | - |
| traffic sign | 72.3955 | - |
| vegetation | 92.6103 | - |
| terrain | 69.9903 | - |
| sky | 94.9167 | - |
| person | 81.5771 | 62.8785 |
| rider | 62.6581 | 39.0113 |
| car | 94.6174 | 87.8882 |
| truck | 49.0696 | 22.0417 |
| bus | 67.1034 | 35.5977 |
| train | 55.3279 | 28.0932 |
| motorcycle | 53.5369 | 35.1325 |
| bicycle | 69.4919 | 53.0499 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.5445 | - |
| nature | 92.2908 | - |
| object | 68.3887 | - |
| sky | 94.9167 | - |
| construction | 91.7939 | - |
| human | 82.5191 | 64.8646 |
| vehicle | 93.8371 | 85.4083 |