Aligning Moments in Time using Video Queries

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kumar, Yogesh, Agarwal, Uday, Gupta, Manish, Mishra, Anand
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911131634761728
author Kumar, Yogesh
Agarwal, Uday
Gupta, Manish
Mishra, Anand
author_facet Kumar, Yogesh
Agarwal, Uday
Gupta, Manish
Mishra, Anand
contents Video-to-video moment retrieval (Vid2VidMR) is the task of localizing unseen events or moments in a target video using a query video. This task poses several challenges, such as the need for semantic frame-level alignment and modeling complex dependencies between query and target videos. To tackle this challenging problem, we introduce MATR (Moment Alignment TRansformer), a transformer-based model designed to capture semantic context as well as the temporal details necessary for precise moment localization. MATR conditions target video representations on query video features using dual-stage sequence alignment that encodes the required correlations and dependencies. These representations are then used to guide foreground/background classification and boundary prediction heads, enabling the model to accurately identify moments in the target video that semantically match with the query video. Additionally, to provide a strong task-specific initialization for MATR, we propose a self-supervised pre-training technique that involves training the model to localize random clips within videos. Extensive experiments demonstrate that MATR achieves notable performance improvements of 13.1% in R@1 and 8.1% in mIoU on an absolute scale compared to state-of-the-art methods on the popular ActivityNet-VRL dataset. Additionally, on our newly proposed dataset, SportsMoments, MATR shows a 14.7% gain in R@1 and a 14.4% gain in mIoU on an absolute scale over strong baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2508_15439
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Aligning Moments in Time using Video Queries
Kumar, Yogesh
Agarwal, Uday
Gupta, Manish
Mishra, Anand
Computer Vision and Pattern Recognition
Video-to-video moment retrieval (Vid2VidMR) is the task of localizing unseen events or moments in a target video using a query video. This task poses several challenges, such as the need for semantic frame-level alignment and modeling complex dependencies between query and target videos. To tackle this challenging problem, we introduce MATR (Moment Alignment TRansformer), a transformer-based model designed to capture semantic context as well as the temporal details necessary for precise moment localization. MATR conditions target video representations on query video features using dual-stage sequence alignment that encodes the required correlations and dependencies. These representations are then used to guide foreground/background classification and boundary prediction heads, enabling the model to accurately identify moments in the target video that semantically match with the query video. Additionally, to provide a strong task-specific initialization for MATR, we propose a self-supervised pre-training technique that involves training the model to localize random clips within videos. Extensive experiments demonstrate that MATR achieves notable performance improvements of 13.1% in R@1 and 8.1% in mIoU on an absolute scale compared to state-of-the-art methods on the popular ActivityNet-VRL dataset. Additionally, on our newly proposed dataset, SportsMoments, MATR shows a 14.7% gain in R@1 and a 14.4% gain in mIoU on an absolute scale over strong baselines.
title Aligning Moments in Time using Video Queries
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.15439