Method Details
Details for method 'ERFNet (pretrained)'
Method overview
| name | ERFNet (pretrained) |
| challenge | pixel-level semantic labeling |
| details | ERFNet pretrained on ImageNet and trained only on the fine train (2975) annotated images |
| publication | ERFNet: Efficient Residual Factorized ConvNet for Real-time Semantic Segmentation Eduardo Romera, Jose M. Alvarez, Luis M. Bergasa and Roberto Arroyo Transactions on Intelligent Transportation Systems (T-ITS) http://www.robesafe.uah.es/personal/eduardo.romera/pdfs/Romera17tits.pdf |
| project page / code | https://github.com/Eromera/erfnet |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | 0.02 s |
| subsampling | 2 |
| submission date | February, 2017 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 69.707 |
| iIoU Classes | 44.0853 |
| IoU Categories | 87.2557 |
| iIoU Categories | 72.6696 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 97.9048 | - |
| sidewalk | 82.116 | - |
| building | 90.678 | - |
| wall | 45.2373 | - |
| fence | 50.4172 | - |
| pole | 58.9734 | - |
| traffic light | 62.6306 | - |
| traffic sign | 68.3736 | - |
| vegetation | 91.8955 | - |
| terrain | 69.4217 | - |
| sky | 94.1698 | - |
| person | 78.5112 | 60.0625 |
| rider | 59.7902 | 34.7242 |
| car | 93.4182 | 86.1115 |
| truck | 52.2916 | 22.5737 |
| bus | 60.8027 | 37.6314 |
| train | 53.7303 | 31.2068 |
| motorcycle | 49.8634 | 29.0017 |
| bicycle | 64.2075 | 51.3709 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.2363 | - |
| nature | 91.5102 | - |
| object | 65.0895 | - |
| sky | 94.1698 | - |
| construction | 90.6326 | - |
| human | 78.899 | 61.2377 |
| vehicle | 92.2522 | 84.1015 |