Video-ColBERT: Contextualized Late Interaction for Text-to-Video Retrieval

Arun Reddy, Alexander Martin, Eugene Yang, Andrew Yates, Kate Sanders, Kenton Murray, Reno Kriz, Celso M. de Melo, Benjamin Van Durme, Rama Chellappa; Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR), 2025, pp. 19691-19701

Abstract

In this work, we tackle the problem of text-to-video retrieval (T2VR). Inspired by the success of late interaction techniques in text-document, text-image, and text-video retrieval, our approach, Video-ColBERT, introduces a simple and efficient mechanism for fine-grained similarity assessment between queries and videos. Video-ColBERT is built upon 3 main components: a fine-grained spatial and temporal token-wise interaction, query and visual expansions, and a dual sigmoid loss during training. We find that this interaction and training paradigm leads to strong individual, yet compatible, representations for encoding video content. These representations lead to increases in performance on common text-to-video retrieval benchmarks compared to other bi-encoder methods.

Related Material

[pdf] [supp]
[bibtex]
@InProceedings{Reddy_2025_CVPR, author = {Reddy, Arun and Martin, Alexander and Yang, Eugene and Yates, Andrew and Sanders, Kate and Murray, Kenton and Kriz, Reno and de Melo, Celso M. and Van Durme, Benjamin and Chellappa, Rama}, title = {Video-ColBERT: Contextualized Late Interaction for Text-to-Video Retrieval}, booktitle = {Proceedings of the Computer Vision and Pattern Recognition Conference (CVPR)}, month = {June}, year = {2025}, pages = {19691-19701} }