Stacked Cross Attention for Image-Text Matching

Benchmark Model Rank Results
cross-modal-retrieval-on-coco-2014SCAN#28Text-to-image R@1: 38.6Text-to-image R@5: 69.3
cross-modal-retrieval-on-flickr30kSCAN#17Image-to-text R@1: 67.4Image-to-text R@5: 90.3
image-retrieval-on-flickr30k-1k-testSCAN i-t#10R@1: 44.0R@5: 74.2R@10: 82.6
image-retrieval-on-photochatSCAN#4R1: 10.4R@10: 37.1R@5: 27.0Sum(R@1,5,10): 74.5