Sketch-Based Image Retrieval on Sketchy
82.7mAP@200TransAgg + UNION
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TransAgg + UNIONBackbone=CLIP-L, # Pairs=5K, Zero-shot=true2025.11 | 82.7 | 79.9 | |
| TransAggBackbone=CLIP-L, # Pairs=5K, Zero-shot=true2025.11 | 79.6 | 75.8 | |
| VaGFeM_TripletBackbone=BLIP-B, Number of Pairs=16K2025.11 | 75.7 | 72.9 | |
| VaGFeMBackbone=BLIP-B, Number of Pairs=16K2025.11 | 74 | 71.9 | |
| TransAgg_FIGROTDBackbone=BLIP-B, Number of Pairs=16K2025.11 | 73.4 | 70.9 | |
| DCDLBackbone=CLIP-B, # Pairs=57K/15K/236K, Zero-shot=true2025.11 | 72.6 | 76.9 | |
| DCDLBackbone=CLIP-B, Number of Pairs=57K/15K/236K2025.11 | 72.6 | 76.9 | |
| CATBackbone=CLIP-B, # Pairs=57K/15K/236K, Zero-shot=true2025.11 | 71.3 | 72.5 | |
| CATBackbone=CLIP-B, Number of Pairs=57K/15K/236K2025.11 | 71.3 | 72.5 | |
| MagicLensBackbone=CLIP-L, # Pairs=36.7M, Zero-shot=true2025.11 | 68.2 | 75.8 | |
| MagicLensBackbone=CLIP-L, Number of Pairs=36.7M2025.11 | 68.2 | 75.8 | |
| IVTBackbone=ViT-B, # Pairs=57K/15K/236K, Zero-shot=true2025.11 | 61.5 | 69.4 | |
| IVTBackbone=ViT-B, Number of Pairs=57K/15K/236K2025.11 | 61.5 | 69.4 | |
| ZSE-SBIRBackbone=ViT-L, # Pairs=57K/15K/236K, Zero-shot=true2025.11 | 52.5 | 62.4 | |
| ZSE-SBIRBackbone=ViT-L, Number of Pairs=57K/15K/236K2025.11 | 52.5 | 62.4 |