Unsupervised open-vocabulary semantic segmentation on COCO Object (val)
37.4mIoUFreeDA (ViT-L)
Evaluation results
| Method | Links | |
|---|---|---|
| FreeDA (ViT-L)PAMR=false2024.04 | 37.4 | |
| TCLPAMR=true, Training Dataset=CC3M+CC12M2024.04 | 31.6 | |
| TCLPAMR=false, Training Dataset=CC3M+CC12M2024.04 | 30.4 | |
| GroupViTPAMR=true, Training Dataset=CC12M+YFCC2024.04 | 27.9 | |
| GroupViTPAMR=false, Training Dataset=CC12M+RedCaps2024.04 | 27.5 | |
| SegCLIPPAMR=false, Training Dataset=CC3M+COCO Captions2024.04 | 26.5 | |
| OVSegmentorPAMR=false, Training Dataset=CC4M2024.04 | 25.1 | |
| ViewCoPAMR=false, Training Dataset=CC12M+YFCC2024.04 | 23.5 | |
| MaskCLIPPAMR=false2024.04 | 20.6 | |
| MaskCLIPPAMR=true2024.04 | 18.9 | |
| ReCoPAMR=false2024.04 | 15.7 |