QVHighlights: Detecting Moments and Highlights in Videos via Natural Language Queries

TL;DR

This paper introduces QVHighlights, a new dataset with over 10,000 videos and annotations for detecting video moments and highlights based on natural language queries, along with a transformer-based baseline model called Moment-DETR.

Contribution

The paper provides the first large-scale dataset for NL-based video moment detection and a novel transformer model that predicts moments directly, improving performance with weakly supervised pretraining.

Findings

Moment-DETR achieves competitive results without human priors.

Weakly supervised pretraining with ASR captions significantly improves performance.

The dataset enables diverse and flexible query-based highlight detection.

Abstract

Detecting customized moments and highlights from videos given natural language (NL) user queries is an important but under-studied topic. One of the challenges in pursuing this direction is the lack of annotated data. To address this issue, we present the Query-based Video Highlights (QVHIGHLIGHTS) dataset. It consists of over 10,000 YouTube videos, covering a wide range of topics, from everyday activities and travel in lifestyle vlog videos to social and political activities in news videos. Each video in the dataset is annotated with: (1) a human-written free-form NL query, (2) relevant moments in the video w.r.t. the query, and (3) five-point scale saliency scores for all query-relevant clips. This comprehensive annotation enables us to develop and evaluate systems that detect relevant moments as well as salient highlights for diverse, flexible user queries. We also present a strong baseline for this task, Moment-DETR, a transformer encoder-decoder model that views moment retrieval as a direct set prediction problem, taking extracted video and query representations as inputs and predicting moment coordinates and saliency scores end-to-end. While our model does not utilize any human prior, we show that it performs competitively when compared to well-engineered architectures. With weakly supervised pretraining using ASR captions, MomentDETR substantially outperforms previous methods. Lastly, we present several ablations and visualizations of Moment-DETR. Data and code is publicly available at https://github.com/jayleicn/moment_detr