Semantic Segmentation on ADE20K 150 classes (test)
29.1mIoUFreeSeg + MAFT
Evaluation results
| Method | Links | |
|---|---|---|
| FreeSeg + MAFTOpen-vocabulary setting=true, Proposal Generator Backbone=ResNet101, CLIP Backbone=ViT-B/16, Module=IP-CLIP Encoder2023.09 | 29.1 | |
| OpenSegOpen-vocabulary setting=true, Additional training data=true2023.09 | 28.6 | |
| OVSegOpen-vocabulary setting=true2023.09 | 24.8 | |
| FreeSegOpen-vocabulary setting=true, Proposal Generator Backbone=ResNet101, CLIP Backbone=ViT-B/162023.09 | 17.9 | |
| LSeg+Open-vocabulary setting=true2023.09 | 13 |