Unsupervised Semantic Segmentation on PASCAL VOC (test)
54.6JSCLIPpy
Evaluation results
| Method | Links | |
|---|---|---|
| CLIPpyTraining Dataset=HQITP-134M, Training Supervision=text, SSP=true2022.10 | 54.6 | |
| GroupViTTraining Dataset=CC-12M & YFCC-100M, Training Supervision=text, SSP=false2022.10 | 51.8 | |
| CLIPpyTraining Dataset=CC-12M, Training Supervision=text, SSP=false2022.10 | 47.5 | |
| COMUSTraining Dataset=ImageNet, Training Supervision=self, SSP=false2022.10 | 47.3 | |
| DINOTraining Dataset=ImageNet, Training Supervision=self, SSP=false2022.10 | 45.9 | |
| GroupViT (our implementation)Training Dataset=CC-12M, Training Supervision=text, SSP=false2022.10 | 42.8 | |
| DINOTraining Dataset=CC-12M & YFCC-100M, Training Supervision=self, SSP=true2022.10 | 41.8 | |
| CLIP (our implementation)Training Dataset=HQITP-134M, Training Supervision=text, SSP=false2022.10 | 38.9 | |
| CLIP (our implementation)Training Dataset=CC-12M, Training Supervision=text, SSP=false2022.10 | 37.3 | |
| DSMTraining Dataset=ImageNet, Training Supervision=self, SSP=false2022.10 | 37.2 | |
| CLIPTraining Dataset=CC-12M & YFCC-100M, Training Supervision=text, SSP=false2022.10 | 28.6 | |
| MoCoTraining Dataset=ImageNet, Training Supervision=self, SSP=false2022.10 | 28.2 | |
| DeiTTraining Dataset=ImageNet, Training Supervision=class, SSP=false2022.10 | 24.6 |