Method Details
Details for method 'Gated-SCNN'
Method overview
| name | Gated-SCNN |
| challenge | pixel-level semantic labeling |
| details | |
| publication | Gated-SCNN: Gated Shape CNNs for Semantic Segmentation Towaki Takikawa, David Acuna, Varun Jampani, Sanja Fidler https://arxiv.org/abs/1907.05740 |
| project page / code | https://nv-tlabs.github.io/GSCNN/ |
| used Cityscapes data | fine annotations |
| used external data | Mapillary |
| runtime | n/a |
| subsampling | no |
| submission date | March, 2019 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 82.7844 |
| iIoU Classes | 64.2946 |
| IoU Categories | 92.3464 |
| iIoU Categories | 82.6642 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.7493 | - |
| sidewalk | 87.4318 | - |
| building | 94.2428 | - |
| wall | 61.9131 | - |
| fence | 64.5662 | - |
| pole | 72.9096 | - |
| traffic light | 79.6054 | - |
| traffic sign | 82.5112 | - |
| vegetation | 94.2912 | - |
| terrain | 74.3352 | - |
| sky | 96.1943 | - |
| person | 88.3366 | 73.7242 |
| rider | 74.1957 | 55.588 |
| car | 96.6236 | 92.3276 |
| truck | 77.151 | 49.1304 |
| bus | 90.1441 | 61.885 |
| train | 87.7142 | 61.37 |
| motorcycle | 72.5896 | 52.9292 |
| bicycle | 79.3984 | 67.4027 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.8427 | - |
| nature | 93.9909 | - |
| object | 78.2516 | - |
| sky | 96.1943 | - |
| construction | 94.4962 | - |
| human | 88.439 | 74.5818 |
| vehicle | 96.2097 | 90.7465 |