1 paper
Zequn Xie, Xin Liu, Boyun Zhang +3
The success of CLIP has driven substantial progress in text-video retrieval. However, current methods often suffer from "blind" feature interaction, where the model struggles to di…