Method Details
Details for method 'M2FANet'
Method overview
| name | M2FANet |
| challenge | pixel-level semantic labeling |
| details | |
| publication | Urban street scene analysis using lightweight multi-level multi-path feature aggregation network Tanmay Singha; Duc-Son Pham; Aneesh Krishna Multiagent and Grid Systems Journal https://content.iospress.com/articles/multiagent-and-grid-systems/mgs210353 |
| project page / code | https://github.com/tanmaysingha/M2FANet |
| used Cityscapes data | fine annotations |
| used external data | |
| runtime | n/a |
| subsampling | no |
| submission date | May, 2021 |
| previous submissions |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 68.3294 |
| iIoU Classes | 38.726 |
| IoU Categories | 86.8694 |
| iIoU Categories | 67.7678 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 97.4247 | - |
| sidewalk | 78.6127 | - |
| building | 90.1355 | - |
| wall | 37.136 | - |
| fence | 42.8683 | - |
| pole | 56.7972 | - |
| traffic light | 63.8496 | - |
| traffic sign | 68.2819 | - |
| vegetation | 92.0695 | - |
| terrain | 66.4723 | - |
| sky | 94.4632 | - |
| person | 79.3377 | 52.6881 |
| rider | 58.7838 | 30.3789 |
| car | 93.6292 | 84.6557 |
| truck | 48.5135 | 18.4318 |
| bus | 62.0532 | 29.8914 |
| train | 53.2208 | 23.421 |
| motorcycle | 48.6754 | 23.6265 |
| bicycle | 65.934 | 46.715 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 96.8642 | - |
| nature | 91.5802 | - |
| object | 63.2465 | - |
| sky | 94.4632 | - |
| construction | 90.0945 | - |
| human | 79.6232 | 53.8094 |
| vehicle | 92.2141 | 81.7262 |