Cluster-Wise Spatio-Temporal Masking for Efficient Video-Language Pretraining

Benchmark Model Rank Results
video-retrieval-on-msr-vttClusterSTM#10text-to-video R@1: 49.7text-to-video R@5: 76.1…
zero-shot-video-retrieval-on-didemoClusterSTM#14text-to-video R@1: 36.5
zero-shot-video-retrieval-on-msr-vttClusterSTM#24text-to-video R@1: 31.2