Method Details
Details for method 'FTFNet'
Method overview
| name | FTFNet |
| challenge | pixel-level semantic labeling |
| details | An Efficient Network Focused on Tiny Feature Maps for Real-Time Semantic Segmentation |
| publication | Anonymous |
| project page / code | |
| used Cityscapes data | fine annotations, coarse annotations |
| used external data | |
| runtime | 0.0088 s 1080Ti |
| subsampling | no |
| submission date | November, 2019 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 72.3533 |
| iIoU Classes | 45.8464 |
| IoU Categories | 87.6069 |
| iIoU Categories | 70.9409 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.2185 | - |
| sidewalk | 83.3144 | - |
| building | 90.9971 | - |
| wall | 47.9422 | - |
| fence | 47.7627 | - |
| pole | 55.3764 | - |
| traffic light | 62.9954 | - |
| traffic sign | 68.6859 | - |
| vegetation | 92.3158 | - |
| terrain | 69.4769 | - |
| sky | 94.7372 | - |
| person | 80.353 | 57.2371 |
| rider | 61.5515 | 35.3966 |
| car | 94.5174 | 86.1306 |
| truck | 62.5282 | 26.4183 |
| bus | 75.8684 | 41.1818 |
| train | 63.0162 | 39.303 |
| motorcycle | 56.5004 | 30.9671 |
| bicycle | 68.5553 | 50.1368 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.4098 | - |
| nature | 91.9839 | - |
| object | 62.6573 | - |
| sky | 94.7372 | - |
| construction | 91.399 | - |
| human | 80.4997 | 58.3741 |
| vehicle | 93.5612 | 83.5077 |