ART: Actor-Related Tubelet for Detecting Complex-shaped Action Tubes

Jiaojiao Zhao, Shuai Liao, Xinyu Li, Bing Shuai, Caren Chen

Keywords: human action recognition, action tube localization

TL;DR: We propose Actor-related Tubelet (ART), which incorporates actor-specific information for generating action tubes with complex shapes.

Abstract: This paper focuses on detecting complex-shaped action tubes in videos. Existing methods are based on the assumption that actor's position changes slightly in short video clips. These methods either oversimplify the shape of action tubes by representing them as cuboids or conjecture that action tubes can be summarized into a set of learnable positional patterns. However, these solutions may be insufficient when actor trajectories become more complex. This limitation arises because these methods rely solely on position information to determine action tubes, lacking the ability to trace the same actor when their movement patterns are intricate. To address this issue, we propose Actor-related Tubelet (ART), which incorporates actor-specific information when generating action tubes. Regardless of the complexity of an actor's trajectory, ART ensures that an action tube consistently tracks the same actor, relying on actor-specific cues rather than solely on positional information. To evaluate the effectiveness of ART in handling complex-shaped action tubes, we introduce a dedicated metric that quantifies tube shape complexity. We conduct experiments on three commonly used tube detection datasets: MultiSports, UCF101-24 and JHMDB51-21. ART presents remarkable improvements on all the datasets.

Primary Area: applications to computer vision, audio, language, and other modalities

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 4837