MomentSeeker: A Task-Oriented Benchmark For Long-Video Moment Retrieval

Ji-Rong Wen (Renmin University of China) · Zheng Liu (Beijing Academy of Artificial Intelligence) · Zhicheng Dou (Renmin University of China) · Junjie Zhou (Nanjing University of Aeronautics and Astronautics) · Bo Zhao (Xi'an Jiaotong University) · Huaying Yuan (Renmin University of China) · Jian Ni (Renmin University of China) · Yueze Wang (Beijing Academy of Artificial Intelligence (BAAI)) · Zhengyang Liang (Singapore Management University) · Zhao Cao (Remin University of China)
action recognitioncausal reasoninggeneration-based approachesimage-conditioned querieslong video understandingmllmsmoment retrievalobject localizationretrieval-based approachestask-specific fine-tuningtext-only queriesvideo-conditioned queries

Accurately locating key moments within long videos is crucial for solving long video understanding (LVU) tasks. However, existing benchmarks are either severely limited in terms of video length and task diversity, or they focus solely on the end-to-end LVU performance, making them inappropriate for evaluating whether key moments can be accurately accessed. To address this challenge, we propose MomentSeeker, a novel benchmark for long-video moment retrieval (LVMR), distinguished by the following features. First, it is created based on long and diverse videos, averaging over 1,200 seconds in duration, and collected from various domains, e.g., movie, anomaly, egocentric, and sports. Second, it covers a variety of real-world scenarios in three levels: global-level, event-level, and object-level, covering common tasks like action recognition, object localization, causal reasoning, etc. Third, it incorporates rich forms of queries, including text-only queries, image-conditioned queries, and video-conditioned queries. On top of MomentSeeker, we conduct comprehensive experiments for both generation-based approaches (directly using MLLMs) and retrieval-based approaches (leveraging video retrievers). Our results reveal the significant challenges in long-video moment retrieval in terms of accuracy and efficiency, despite improvements from the latest long-video MLLMs and task-specific fine-tuning. We have publicly released MomentSeeker to facilitate future research in this area.