Method Details
Details for method 'FSFFNet'
Method overview
| name | FSFFNet |
| challenge | pixel-level semantic labeling |
| details | Feature Scaling Feature Fusion Network |
| publication | A Lightweight Multi-scale Feature Fusion Network for Real-Time Semantic Segmentation Tanmay Singha, Duc-Son Pham, Aneesh Krishna, Tom Gedeon International Conference on Neural Information Processing 2021 https://link.springer.com/chapter/10.1007/978-3-030-92270-2_17 |
| project page / code | https://github.com/tanmaysingha/FSFFNet |
| used Cityscapes data | fine annotations, coarse annotations |
| used external data | |
| runtime | n/a |
| subsampling | no |
| submission date | June, 2021 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 69.38 |
| iIoU Classes | 40.3635 |
| IoU Categories | 87.068 |
| iIoU Categories | 68.472 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 97.4411 | - |
| sidewalk | 78.4959 | - |
| building | 90.7274 | - |
| wall | 41.7536 | - |
| fence | 46.0893 | - |
| pole | 57.8297 | - |
| traffic light | 65.3265 | - |
| traffic sign | 68.5013 | - |
| vegetation | 92.0029 | - |
| terrain | 63.9974 | - |
| sky | 94.4019 | - |
| person | 79.1799 | 53.9591 |
| rider | 56.992 | 31.5266 |
| car | 93.9136 | 84.9889 |
| truck | 55.3569 | 22.2772 |
| bus | 65.689 | 32.0117 |
| train | 54.3536 | 26.5156 |
| motorcycle | 50.3915 | 26.0818 |
| bicycle | 65.7774 | 45.5468 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 96.7782 | - |
| nature | 91.4755 | - |
| object | 64.1045 | - |
| sky | 94.4019 | - |
| construction | 90.193 | - |
| human | 79.7861 | 55.1737 |
| vehicle | 92.737 | 81.7704 |