Generalized Category Discovery on ImageNet-100 (test)
86.5Accuracy (All)SIEFormer
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SIEFormerBase Framework=SimGCD, Backbone=ViT-B/16, Pre-training=DINO2026.02 | 86.5 | 94.5 | 82.6 | — | — | |
| SIEFormerBase Framework=SPTNet, Backbone=ViT-B/16, Pre-training=DINO2026.02 | 86.3 | 94.2 | 82.3 | — | — | |
| SPTNetBackbone=ViT-B/16, Pre-training=DINO2026.02 | 85.4 | 93.2 | 81.4 | — | — | |
| PromptCALsetting=Inductive GCD, backbone=ViT2022.12 | 84.8 | 94.4 | 75.2 | 94.4 | 75.3 | |
| CMSBackbone=ViT-B/16, Pre-training=DINO2026.02 | 84.7 | 95.6 | 79.2 | — | — | |
| VLM-NCDImage Encoder=CLIP, Knowledge Database=Yes2025.12 | 84 | 95.5 | 78.2 | — | — | |
| ProtoGCDBackbone=ViT-B/16, Pre-training=DINO2026.02 | 84 | 92.2 | 79.9 | — | — | |
| PromptCALBackbone=ViT-B/16, Pre-training=DINO2026.02 | 83.1 | 92.7 | 78.3 | — | — | |
| SimGCDBackbone=ViT-B/16, Pre-training=DINO2026.02 | 83 | 93.1 | 77.9 | — | — | |
| ORCA (DINO)setting=Inductive GCD, backbone=ViT2022.12 | 81.3 | 94.5 | 68 | 94.5 | 71.1 | |
| DCCLBackbone=ViT-B/16, Pre-training=DINO2026.02 | 80.5 | 90.5 | 76.2 | — | — | |
| SIEFormerBase Framework=GCD, Backbone=ViT-B/16, Pre-training=DINO2026.02 | 79.9 | 91 | 74.3 | — | — | |
| GCDsetting=Inductive GCD, backbone=ViT2022.12 | 79.7 | 92.7 | 66.7 | 92.7 | 66.9 | |
| XCon2025.12 | 77.6 | 93.5 | 69.7 | — | — | |
| GCD2025.12 | 74.1 | 89.8 | 66.3 | — | — | |
| GCDBackbone=ViT-B/16, Pre-training=DINO2026.02 | 74.1 | 89.8 | 66.3 | — | — | |
| ORCABackbone=ViT-B/16, Pre-training=DINO2026.02 | 73.5 | 92.6 | 63.9 | — | — | |
| UNO2025.12 | 70.3 | 95 | 57.9 | — | — | |
| UNO+Backbone=ViT-B/16, Pre-training=DINO2026.02 | 70.3 | 95 | 57.9 | — | — | |
| RankStats2025.12 | 37.1 | 61.6 | 24.8 | — | — | |
| RankStats+Backbone=ViT-B/16, Pre-training=DINO2026.02 | 37.1 | 61.6 | 24.8 | — | — |