Image Clustering on ImageNet-1K
90.98NMISAGL
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SAGL2026.05 | 90.98 | 79.34 | 70.5 | |
| MSRL2026.05 | 89.17 | 74.65 | 65.14 | |
| TURTLE2026.05 | 88.12 | 71.98 | 61.72 | |
| DCRBackbone=SigLIP ViT-SO@224 (S-1), Zero-shot=true2026.03 | 88 | 65 | 55 | |
| SRLF2026.05 | 85.07 | 69.41 | 56.09 | |
| Original CLIPBackbone=SigLIP ViT-SO@224 (S-1), Zero-shot=true2026.03 | 85 | 63 | 52 | |
| DIVABackbone=SigLIP ViT-SO@224 (S-1), Zero-shot=true2026.03 | 85 | 63 | 52 | |
| MIM2026.05 | 84.72 | 63.24 | 40.18 | |
| DCRBackbone=OpenAI CLIP ViT-L@224 (O-1), Zero-shot=true2026.03 | 84 | 66 | 45 | |
| DCRBackbone=MetaCLIP ViT-L@224 (M-1), Zero-shot=true2026.03 | 84 | 69 | 45 | |
| GSEC2026.05 | 81.3 | 62.5 | 52.8 | |
| CLIPClustering setting=zero-shot2023.10 | 81 | 63.6 | 45.4 | |
| Original CLIPBackbone=MetaCLIP ViT-L@224 (M-1), Zero-shot=true2026.03 | 81 | 56 | 44 | |
| DIVABackbone=MetaCLIP ViT-L@224 (M-1), Zero-shot=true2026.03 | 81 | 56 | 44 | |
| CLIPEvaluation Protocol=zero-shot2025.10 | 81 | 63.6 | 45.4 | |
| TACClustering setting=default2023.10 | 79.9 | 58.2 | 43.5 | |
| GradNorm2025.10 | 79.2 | 52.6 | 39.1 | |
| GradNorm2026.05 | 79.2 | 52.6 | 39.1 | |
| Original CLIPBackbone=OpenAI CLIP ViT-L@224 (O-1), Zero-shot=true2026.03 | 79 | 53 | 41 | |
| DIVABackbone=OpenAI CLIP ViT-L@224 (O-1), Zero-shot=true2026.03 | 79 | 52 | 40 | |
| TACClustering setting=no train2023.10 | 77.8 | 48.9 | 36.4 | |
| TAC2025.10 | 77.8 | 48.9 | 36.4 | |
| TAC2026.05 | 77.8 | 48.9 | 36.4 | |
| SICClustering setting=default2023.10 | 77.2 | 47 | 34.3 | |
| SIC2025.10 | 77.2 | 47 | 34.3 | |
| SIC2026.05 | 77.2 | 47 | 34.3 | |
| GenHancerBackbone=SigLIP ViT-SO@224 (S-1), Zero-shot=true2026.03 | 77 | 48 | 35 | |
| SCANClustering setting=default2023.10 | 74.7 | 44.7 | 32.4 | |
| SCAN2025.10 | 74.7 | 44.7 | 32.4 | |
| un2CLIPBackbone=OpenAI CLIP ViT-L@224 (O-1), Zero-shot=true2026.03 | 73 | 41 | 28 | |
| CLIPClustering setting=k-means2023.10 | 72.3 | 38.9 | 27.1 | |
| CLIPEvaluation Protocol=k-means2025.10 | 72.3 | 38.9 | 27.1 | |
| CLIPClustering Algorithm=k-means2026.05 | 72.3 | 38.9 | 27.1 | |
| PRR2026.05 | 68.53 | 72.97 | 58.92 | |
| GenHancerBackbone=OpenAI CLIP ViT-L@224 (O-1), Zero-shot=true2026.03 | 64 | 26 | 14 | |
| GenHancerBackbone=MetaCLIP ViT-L@224 (M-1), Zero-shot=true2026.03 | 64 | 27 | 15 |