Method Details
Details for method 'Dense Prediction with Attentive Feature aggregation'
Method overview
| name | Dense Prediction with Attentive Feature aggregation |
| challenge | pixel-level semantic labeling |
| details | We propose Attentive Feature Aggregation (AFA) to exploit both spatial and channel information for semantic segmentation and boundary detection. |
| publication | Dense Prediction with Attentive Feature Aggregation Yung-Hsu Yang, Thomas E. Huang, Min Sun, Samuel Rota Bulò, Peter Kontschieder, Fisher Yu WACV 2023 https://arxiv.org/abs/2111.00770 |
| project page / code | https://www.vis.xyz/pub/dla-afa/ |
| used Cityscapes data | fine annotations, coarse annotations |
| used external data | ImageNet |
| runtime | n/a |
| subsampling | no |
| submission date | January, 2023 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 83.5806 |
| iIoU Classes | 65.1356 |
| IoU Categories | 92.504 |
| iIoU Categories | 82.21 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.8689 | - |
| sidewalk | 88.6062 | - |
| building | 94.3646 | - |
| wall | 68.3507 | - |
| fence | 64.8711 | - |
| pole | 72.6637 | - |
| traffic light | 80.688 | - |
| traffic sign | 83.3882 | - |
| vegetation | 94.1919 | - |
| terrain | 74.5032 | - |
| sky | 96.133 | - |
| person | 89.1327 | 73.5188 |
| rider | 76.8841 | 57.1693 |
| car | 96.6786 | 91.0812 |
| truck | 76.4234 | 51.0921 |
| bus | 90.4012 | 63.9609 |
| train | 89.0401 | 57.3074 |
| motorcycle | 73.2579 | 57.9308 |
| bicycle | 79.5838 | 69.0247 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.8905 | - |
| nature | 94.0155 | - |
| object | 78.3673 | - |
| sky | 96.133 | - |
| construction | 94.5283 | - |
| human | 89.2818 | 74.6157 |
| vehicle | 96.3119 | 89.8044 |