Method Details
Details for method 'VLTSeg '
Method overview
| name | VLTSeg |
| challenge | pixel-level semantic labeling |
| details | |
| publication | VLTSeg: Simple Transfer of CLIP-Based Vision-Language Representations for Domain Generalized Semantic Segmentation Christoph Hümmer, Manuel Schwonberg, Liangwei Zhou, Hu Cao, Alois Knoll, Hanno Gottschalk https://arxiv.org/abs/2312.02021 |
| project page / code | |
| used Cityscapes data | fine annotations |
| used external data | Mapillary, Laion-2B Merged |
| runtime | n/a |
| subsampling | no |
| submission date | January, 2024 |
| previous submissions | 1 |
Average results
| Metric | Value |
|---|---|
| IoU Classes | 86.3781 |
| iIoU Classes | 73.2919 |
| IoU Categories | 93.1434 |
| iIoU Categories | 85.3231 |
Class results
| Class | IoU | iIoU |
|---|---|---|
| road | 98.9082 | - |
| sidewalk | 88.9754 | - |
| building | 94.7989 | - |
| wall | 70.9125 | - |
| fence | 72.7899 | - |
| pole | 75.5606 | - |
| traffic light | 81.3257 | - |
| traffic sign | 84.6456 | - |
| vegetation | 94.5457 | - |
| terrain | 75.6813 | - |
| sky | 96.276 | - |
| person | 90.5664 | 79.3509 |
| rider | 81.4252 | 66.48 |
| car | 96.8312 | 91.3832 |
| truck | 84.2392 | 63.0101 |
| bus | 95.4353 | 73.3479 |
| train | 93.176 | 71.5117 |
| motorcycle | 82.3596 | 67.8939 |
| bicycle | 82.7314 | 73.3573 |
Category results
| Category | IoU | iIoU |
|---|---|---|
| flat | 98.8824 | - |
| nature | 94.227 | - |
| object | 80.5268 | - |
| sky | 96.276 | - |
| construction | 95.0417 | - |
| human | 90.5194 | 79.9942 |
| vehicle | 96.5305 | 90.652 |