COOT: Cooperative Hierarchical Transformer for Video-Text Representation Learning

Benchmark Model Rank Results
video-captioning-on-activitynet-captionsCOOT (ae-test split) - Only Appearance features#3BLEU4: 10.85BLEU-3: 17.43CIDEr: 28.19ROUGE-L: 31.45…
video-captioning-on-youcook2COOT#6BLEU-4: 11.30BLEU-3: 17.97CIDEr: 0.57ROUGE-L: 37.94…
video-retrieval-on-youcook2COOT#8text-to-video R@1: 16.7text-to-video R@10: 52.3…