ShadowPunch: fast actions spotting benchmark

Aleksandr Simonyan, Nikolay Falaleev

Keywords: Action Spotting, Pose Estimation, Dataset, Boxing, Sports, Video Understanding

TL;DR: A new open video dataset for high-frame-rate shadow boxing action spotting has been introduced as a benchmark for two event spotting methods: direct prediction from image data and a staged approach that utilizes intermediate pose estimation.

Abstract: We introduce an open dataset for video event spotting focused on fast-paced events in shadowboxing videos captured at high frame rates. The dataset features accurate frame-level annotations for diverse punch types alongside pose keypoint annotations, enabling the development of robust event recognition models. This work presents a novel benchmark exploring two distinct approaches to event spotting: direct prediction from image data and a staged approach involving intermediate pose estimation followed by event detection based on the detected keypoints. We provide baseline neural network solutions incorporating temporal information for both tracks, facilitating comparative analysis of these methodologies. This shadowboxing dataset advances the field of automatic sports analysis and contributes to the broader understanding of video events recognition.

Primary Area: datasets and benchmarks

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: 13079