Semantic Correspondence on AP-10K cross-family
83.4PCK@0.10MARCO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MARCOSupervision=Supervised, DINOv2 Encoder=true, SD Encoder=false, Uses object masks at training=true2026.04 | 83.4 | 28.5 | 71.1 | |
| MARCOSupervision=Supervised, DINOv2 Encoder=true, SD Encoder=false2026.04 | 82.5 | 27.4 | 69.9 | |
| SimpleMatchInput size=448x4482026.01 | 81.3 | 25 | 68.6 | |
| GeoAware-SC2026.01 | 78.5 | 18.3 | 63.2 | |
| Geo-SCSupervision=Supervised, DINOv2 Encoder=true, SD Encoder=true, Uses object masks at inference=true2026.04 | 78.5 | 18.3 | 63.2 | |
| GECOSupervision=Supervised, DINOv2 Encoder=false, SD Encoder=true, Uses object masks at training=true2026.04 | 76.6 | 14.5 | 60.4 | |
| SD+DINO(S)2026.01 | 65.8 | 6.9 | 46.2 | |
| SD+DINOv2Supervision=Supervised, DINOv2 Encoder=true, SD Encoder=true2026.04 | 65.8 | 6.9 | 46.2 | |
| SD+DINO (S)Supervision=Supervised2026.07 | 65.8 | 6.9 | 46.2 | |
| DIY-SCSupervision=Unsupervised / Weakly Supervised, DINOv2 Encoder=true, SD Encoder=true, Uses object masks at training=true2026.04 | 57.8 | — | — | |
| MREPSupervision=Unsupervised2026.07 | 49.6 | 5.9 | 35 | |
| DiTFSupervision=Unsupervised2026.07 | 48.9 | 5.7 | 34.2 | |
| SD+DINOv2Supervision=Unsupervised / Weakly Supervised, DINOv2 Encoder=true, SD Encoder=true2026.04 | 48.3 | 5.2 | 30.8 | |
| DHF2026.01 | 47.8 | 5 | 32.7 | |
| DHFSupervision=Supervised, DINOv2 Encoder=false, SD Encoder=true2026.04 | 47.8 | 5 | 32.7 | |
| DHFSupervision=Supervised2026.07 | 47.8 | 5 | 32.7 | |
| DINOv2+NNSupervision=Unsupervised / Weakly Supervised, DINOv2 Encoder=true, SD Encoder=false2026.04 | 47.4 | 4.4 | 29.4 | |
| DINOv2+NNSupervision=Unsupervised2026.07 | 47.4 | 4.4 | 29.4 | |
| DIFTSupervision=Unsupervised / Weakly Supervised, DINOv2 Encoder=false, SD Encoder=true2026.04 | 35 | 3.7 | 22.4 | |
| DIFTSupervision=Unsupervised2026.07 | 35 | 3.7 | 22.4 | |
| w/o MREPSupervision=Unsupervised2026.07 | 22.1 | 2.3 | 14.2 |