Method Details
Details for method 'DGCNet'
Method overview
| name | DGCNet |
| challenge | pixel-level semantic labeling |
| details | We propose Dual Graph Convolutional Network (DGCNet) models the global context of the input feature by modelling two orthogonal graphs in a single framework. (Joint work: University of Oxford, Peking University and DeepMotion AI Research) |
| publication | Dual Graph Convolutional Network for Semantic Segmentation Li Zhang*, Xiangtai Li*, Anurag Arnab, Kuiyuan Yang, Yunhai Tong, Philip H.S. Torr BMVC 2019 |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | ImageNet |
| runtime | n/a |
| subsampling | no |
| submission date | April, 2019 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 81.9622 |
| iIoU Classes | 61.7449 |
| IoU Categories | 91.8358 |
| iIoU Categories | 81.076 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.7394 | - |
| sidewalk | 87.4209 | - |
| building | 93.9256 | - |
| wall | 62.3839 | - |
| fence | 63.3652 | - |
| pole | 70.9209 | - |
| traffic light | 78.6795 | - |
| traffic sign | 81.3203 | - |
| vegetation | 93.9546 | - |
| terrain | 73.3012 | - |
| sky | 95.8296 | - |
| person | 87.8335 | 71.9745 |
| rider | 73.7451 | 53.9179 |
| car | 96.3744 | 91.2109 |
| truck | 75.9963 | 47.1444 |
| bus | 91.6062 | 57.5634 |
| train | 81.6435 | 53.9978 |
| motorcycle | 71.5419 | 51.9066 |
| bicycle | 78.7003 | 66.2438 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.7579 | - |
| nature | 93.6424 | - |
| object | 76.5851 | - |
| sky | 95.8296 | - |
| construction | 94.1302 | - |
| human | 87.9952 | 72.9057 |
| vehicle | 95.9103 | 89.2462 |