Fine-grained object category discovery on Stanford Cars (test)
78AccuracyFlipClass
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| FlipClassBackbone=DINOv22024.09 | 78 | — | 88 | 73.2 | — | — | |
| μGCDBackbone=DINOv22024.09 | 76.1 | — | 91 | 68.9 | — | — | |
| SimGCDBackbone=DINOv22024.09 | 71.5 | — | 81.9 | 64.6 | — | — | |
| GCDBackbone=DINOv22024.09 | 65.7 | — | 67.8 | 64.7 | — | — | |
| FlipClassBackbone=DINO2024.09 | 63.1 | — | 81.7 | 53.8 | — | — | |
| SPTNetBackbone=DINO2024.09 | 59 | — | 79.2 | 49.3 | — | — | |
| CMSBackbone=DINO2024.09 | 56.9 | — | 76.1 | 47.6 | — | — | |
| μGCDBackbone=DINO2024.09 | 56.5 | — | 68.1 | 50.9 | — | — | |
| AMENDBackbone=DINO2024.09 | 56.4 | — | 73.3 | 48.2 | — | — | |
| InfoSieveBackbone=DINO2024.09 | 55.7 | — | 74.8 | 46.4 | — | — | |
| TIDABackbone=DINO2024.09 | 54.7 | — | 72.3 | 46.2 | — | — | |
| GCABackbone=DINO2024.09 | 54.4 | — | 72.1 | 45.8 | — | — | |
| SimGCDBackbone=ViT-B/16, Pre-trained=DINO2022.11 | 53.8 | — | 71.9 | 45 | — | — | |
| SimGCDBackbone=DINO2024.09 | 53.8 | — | 71.9 | 45 | — | — | |
| PromptCALStage=2nd2022.12 | 50.2 | — | 70.1 | 40.6 | — | — | |
| PCALBackbone=DINO2024.09 | 50.2 | — | 70.1 | 40.6 | — | — | |
| AdaptGCDBackbone=DINO2024.09 | 48.4 | — | 57.7 | 39.3 | — | — | |
| CiPRBackbone=DINO2024.09 | 47 | — | 61.5 | 40.1 | — | — | |
| PromptCALStage=1st2022.12 | 42.6 | — | 44.6 | 44.5 | — | — | |
| XConBackbone=DINO2024.09 | 40.5 | — | 58.8 | 31.7 | — | — | |
| GCD2022.12 | 39 | — | 57.6 | 29.9 | — | — | |
| GCDBackbone=ViT-B/16, Pre-trained=DINO2022.11 | 39 | — | 57.6 | 29.9 | — | — | |
| GCDBackbone=DINO2024.09 | 39 | — | 57.6 | 29.9 | — | — | |
| OpenLDN-UDABackbone=ResNet-18, Labeled Ratio=50%, Class Split=50% known / 50% novel2022.07 | 38.7 | — | — | — | — | — | |
| UNOadapted=true2022.12 | 35.5 | — | 70.5 | 18.6 | — | — | |
| UNO+Backbone=ViT-B/16, Pre-trained=DINO2022.11 | 35.5 | — | 70.5 | 18.6 | — | — | |
| ORCAadapted=true2022.12 | 31.9 | — | 42.2 | 26.9 | — | — | |
| RankStatsadapted=true2022.12 | 28.3 | — | 61.8 | 12.1 | — | — | |
| RS+Backbone=ViT-B/16, Pre-trained=DINO2022.11 | 28.3 | — | 61.8 | 12.1 | — | — | |
| ORCABackbone=ViT-B/16, Pre-trained=DINO2022.11 | 23.5 | — | 50.1 | 10.7 | — | — | |
| KMeans2022.12 | 12.8 | — | 10.6 | 13.8 | — | — | |
| k-meansBackbone=ViT-B/16, Pre-trained=DINO2022.11 | 12.8 | — | 10.6 | 13.8 | — | — | |
| ORCABackbone=ResNet-18, Labeled Ratio=50%, Class Split=50% known / 50% novel2022.07 | 9.6 | — | — | — | — | — | |
| FineGAN2018.11 | 7.8 | 35.4 | — | — | — | — | |
| DEPICT2018.11 | 6.3 | 32.9 | — | — | — | — | |
| DEPICT-LargeCapacity=Large (Double filters)2018.11 | 6.2 | 33 | — | — | — | — | |
| JULE-ResNet-50Backbone=ResNet-502018.11 | 5 | 23.7 | — | — | — | — | |
| JULE2018.11 | 4.6 | 23.2 | — | — | — | — | |
| PHE2024.10 | — | — | — | 16.8 | 31.3 | 61.9 | |
| RankStat2024.10 | — | — | — | 9.7 | 18.6 | 36.9 | |
| SLC2024.10 | — | — | — | 13.6 | 24 | 45.8 | |
| SMILE2024.10 | — | — | — | 16.3 | 26.2 | 46.7 | |
| WTA2024.10 | — | — | — | 10.6 | 20 | 38.8 |