Image-Text Retrieval on Crisscrossed Captions (CxC)
58.7R@1VACSR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VACSRNoise Ratio=20%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 58.7 | 539 | |
| VACSRNoise Ratio=50%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 57.2 | 534.2 | |
| VSE∞Noise Ratio=20%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 53.6 | 518.6 | |
| PCME++Noise Ratio=20%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 52.2 | 524.6 | |
| PCMENoise Ratio=20%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 50.6 | 520.7 | |
| NCRNoise Ratio=20%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 50.6 | 518.6 | |
| PCME++Noise Ratio=50%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 47.4 | 511 | |
| NCRNoise Ratio=50%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 47.3 | 508.5 | |
| PCMENoise Ratio=50%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 46.3 | 508 | |
| VSE∞Noise Ratio=50%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 20.7 | 394.1 | |
| DAANoise Ratio=20%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 7 | 212.8 | |
| DAANoise Ratio=50%, Backbone=ViT-B/32, Retrieval Direction=Average2026.05 | 0.3 | 20.9 |