Generalized Category Discovery on SSB Average
86Accuracy (All)Hyp-SelEx
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Hyp-SelExVenue=CVPR’25, Backbone=DINOv22025.09 | 86 | 86.9 | 85.5 | |
| PartCo-SelExVenue=Ours, Backbone=DINOv22025.09 | 85.5 | 86.6 | 84.8 | |
| NCGCD-SelExVenue=NeurIPS’25, Backbone=DINOv22025.09 | 84 | 88.2 | 83.7 | |
| SelExVenue=ECCV’24, Backbone=DINOv22025.09 | 83.1 | 87 | 81.3 | |
| ConGCD-SelExVenue=ICCV’25, Backbone=DINOv22025.09 | 82.6 | 87.9 | 80 | |
| PartCo-SelExVenue=Ours, Backbone=DINOv32025.09 | 81.6 | 86.2 | 79.3 | |
| PartCo-FlipClassVenue=Ours, Backbone=DINOv22025.09 | 80.9 | 85.8 | 78.6 | |
| PartCo-SPTNetVenue=Ours, Backbone=DINOv22025.09 | 80.4 | 84 | 78.1 | |
| Hyp-SimGCDVenue=CVPR’25, Backbone=DINOv22025.09 | 78.8 | 78 | 79.3 | |
| PartCo-SimGCDVenue=Ours, Backbone=DINOv22025.09 | 78.8 | 84.9 | 75.9 | |
| SelExVenue=ECCV’24, Backbone=DINOv32025.09 | 78.7 | 81.9 | 77.3 | |
| PartCo-SimGCDVenue=Ours, Backbone=DINOv32025.09 | 77.5 | 83.8 | 74.5 | |
| RLCDVenue=ICML’25, Backbone=DINOv22025.09 | 76.9 | 82.9 | 74 | |
| SEALVenue=NeurIPS’25, Backbone=DINOv22025.09 | 76.3 | 80.1 | 74.5 | |
| FlipClassVenue=NeurIPS’24, Backbone=DINOv22025.09 | 76.1 | 81.3 | 73.6 | |
| PartGCDVenue=TMM, Backbone=DINOv22025.09 | 75.6 | 81.6 | 72.7 | |
| MOSVenue=CVPR’25, Backbone=DINOv22025.09 | 75.4 | 83.3 | 71.6 | |
| DebGCDVenue=ICLR’25, Backbone=DINOv22025.09 | 74.9 | 81.5 | 71.7 | |
| ProtoGCDVenue=TPAMI, Backbone=DINOv22025.09 | 74.8 | 82.7 | 71 | |
| SimGCDVenue=ICCV’23, Backbone=DINOv32025.09 | 73.7 | 82.5 | 69.4 | |
| AFGCDVenue=ICCV’25, Backbone=DINOv22025.09 | 73.4 | 79.8 | 70.1 | |
| APLVenue=CVPR’25, Backbone=DINOv22025.09 | 72.4 | 80.3 | 68.8 | |
| SPTNetVenue=ICLR’24, Backbone=DINOv22025.09 | 72.2 | 78.9 | 67.6 | |
| µGCDVenue=NeurIPS’23, Backbone=DINOv22025.09 | 72.1 | 78.6 | 69 | |
| GETEncoder=Image & Text, Backbone=CLIP2025.09 | 71.5 | 74.8 | 69.8 | |
| PartCo-SimGCD (Ours)Encoder=Image, Backbone=CLIP2025.09 | 70.5 | 77.6 | 67 | |
| SimGCDVenue=ICCV’23, Backbone=DINOv22025.09 | 69 | 76.6 | 65.3 | |
| SimGCDEncoder=Image, Backbone=CLIP2025.09 | 65.3 | 72.8 | 61.7 | |
| GCDVenue=CVPR’22, Backbone=DINOv22025.09 | 64.3 | 62.3 | 65.4 | |
| CLIP-GCDEncoder=Image & Text, Backbone=CLIP2025.09 | 61.1 | 74 | 54.8 | |
| k-meansBackbone=DINOv32025.09 | 56.4 | 50.3 | 59.4 | |
| GCDEncoder=Image, Backbone=CLIP2025.09 | 56 | 61.8 | 53.1 | |
| k-meansBackbone=DINOv22025.09 | 38.6 | 34 | 40 |