Method Details
Details for method 'Edgenet'
Method overview
| name | Edgenet |
| challenge | pixel-level semantic labeling |
| details | A lightweight semantic segmentation network combined with edge information and channel-wise attention mechanism. |
| publication | Anonymous |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | 0.03 s TitanX(Maxwell) PyTorch |
| subsampling | 2 |
| submission date | June, 2018 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 71.0106 |
| iIoU Classes | 46.5689 |
| IoU Categories | 88.4827 |
| iIoU Categories | 75.0044 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.1151 | - |
| sidewalk | 83.1432 | - |
| building | 91.5525 | - |
| wall | 45.3795 | - |
| fence | 50.5943 | - |
| pole | 62.6037 | - |
| traffic light | 67.2262 | - |
| traffic sign | 71.4067 | - |
| vegetation | 92.4392 | - |
| terrain | 69.746 | - |
| sky | 94.9343 | - |
| person | 80.3713 | 62.5233 |
| rider | 61.1002 | 38.0991 |
| car | 94.2508 | 88.3789 |
| truck | 49.9648 | 25.4088 |
| bus | 60.9317 | 38.2996 |
| train | 52.5238 | 32.982 |
| motorcycle | 55.2504 | 34.3692 |
| bicycle | 67.6678 | 52.4902 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.4044 | - |
| nature | 92.081 | - |
| object | 68.4808 | - |
| sky | 94.9343 | - |
| construction | 91.4799 | - |
| human | 80.9195 | 63.9222 |
| vehicle | 93.0792 | 86.0866 |