Open Vocabulary Semantic Segmentation on PASCAL VOC VOC20 without background 2012
88.7mIoUTrident
Evaluation results
| Method | Links | ||
|---|---|---|---|
| TridentTrain=false, Backbone=ViT-H/142024.11 | 88.7 | 48.6 | |
| DeCLIP2025.05 | 85 | — | |
| TridentTrain=false, Backbone=ViT-B/162024.11 | 84.5 | 45.8 | |
| ProxyCLIPTrain=false, Backbone=ViT-H/142024.11 | 83.3 | 44.4 | |
| CLIP-DINOiserTrain=true2024.11 | 80.9 | 40.3 | |
| OVDiffTrain=false, Backbone=ViT-B/162024.11 | 80.9 | 37.8 | |
| ClearCLIP2025.05 | 80.9 | — | |
| CLIP-DINOiser2025.05 | 80.9 | — | |
| SCLIPTrain=false, Backbone=ViT-B/162024.11 | 80.4 | 38.2 | |
| SCLIP2025.05 | 80.4 | — | |
| ProxyCLIPTrain=false, Backbone=ViT-B/162024.11 | 80.3 | 42.3 | |
| NACLIPTrain=false, Backbone=ViT-B/162024.11 | 79.7 | 39.4 | |
| GroupViT2025.05 | 79.7 | — | |
| TCLTrain=true2024.11 | 77.5 | 33.9 | |
| TCL2025.05 | 77.5 | — | |
| MaskCLIPTrain=false, Backbone=ViT-B/162024.11 | 74.9 | 30.3 | |
| MaskCLIP2025.05 | 74.9 | — | |
| SCLIPTrain=false, Backbone=ViT-H/142024.11 | 67.5 | 29.1 | |
| ReCoTrain=false, Backbone=ViT-B/162024.11 | 57.7 | 23.5 | |
| ReCo2025.05 | 57.7 | — | |
| CLIP2025.05 | 49.4 | — | |
| CLIPTrain=false, Backbone=ViT-B/162024.11 | 41.9 | 11.7 | |
| MaskCLIPTrain=false, Backbone=ViT-H/142024.11 | 41.7 | 19.3 | |
| OpenCLIPTrain=false, Backbone=ViT-H/142024.11 | 21.7 | 7.1 |