Creating a Novel Attention-Enhanced Framework for Video-Based Action Quality Assessment

Gong, Wenhui; Li, Wei; Hu, Huosheng; Song, Zhijun; Zeng, Zhiqiang; Sun, Jinhua; Song, Yuping

doi:10.3390/sci7020054

Open AccessArticle

Creating a Novel Attention-Enhanced Framework for Video-Based Action Quality Assessment

by

Wenhui Gong

¹

,

Wei Li

^1,*

,

Huosheng Hu

^2,*

,

Zhijun Song

³

,

Zhiqiang Zeng

¹

,

Jinhua Sun

¹

and

Yuping Song

⁴

¹

School of Computer and Information Engineering, Xiamen University of Technology, Xiamen 361024, China

²

School of Computer Science and Electronic Engineering, University of Essex, Colchester CO4 3SQ, UK

³

School of Physics and Information Engineering, Jiangsu Second Normal University, Nanjing 210013, China

⁴

School of Mathematical Sciences, Xiamen University, Xiamen 361005, China

^*

Authors to whom correspondence should be addressed.

Sci 2025, 7(2), 54; https://doi.org/10.3390/sci7020054

Submission received: 12 February 2025 / Revised: 25 April 2025 / Accepted: 28 April 2025 / Published: 6 May 2025

Download

Browse Figures

Versions Notes

Abstract

Action Quality Assessment (AQA)—the task of evaluating how well human actions are performed—is essential in domains such as sports and medicine. Existing AQA methods typically rely on score regression following feature extraction but often neglect the ambiguity inherent in extracted features. In this work, we introduce a novel AQA framework that incorporates a modified attention module to better capture relevant information. Our approach segments video data into clips, extracts features using the I3D network, and applies attention mechanisms to highlight salient features while suppressing irrelevant ones. To assess feature quality, we employ score distribution regression and propose an uncertainty-aware score distribution learning strategy that models features as Gaussian distributions. We further leverage Variational Autoencoders (VAEs) to capture complex latent representations and quantify uncertainty. Extensive experiments on the MTL-AQA and JIGSAWS datasets demonstrate the effectiveness and robustness of our proposed method.

Keywords: attention mechanism; I3D network; feature extraction; video action quality assessment

Share and Cite

MDPI and ACS Style

Gong, W.; Li, W.; Hu, H.; Song, Z.; Zeng, Z.; Sun, J.; Song, Y. Creating a Novel Attention-Enhanced Framework for Video-Based Action Quality Assessment. Sci 2025, 7, 54. https://doi.org/10.3390/sci7020054

AMA Style

Gong W, Li W, Hu H, Song Z, Zeng Z, Sun J, Song Y. Creating a Novel Attention-Enhanced Framework for Video-Based Action Quality Assessment. Sci. 2025; 7(2):54. https://doi.org/10.3390/sci7020054

Chicago/Turabian Style

Gong, Wenhui, Wei Li, Huosheng Hu, Zhijun Song, Zhiqiang Zeng, Jinhua Sun, and Yuping Song. 2025. "Creating a Novel Attention-Enhanced Framework for Video-Based Action Quality Assessment" Sci 7, no. 2: 54. https://doi.org/10.3390/sci7020054

APA Style

Gong, W., Li, W., Hu, H., Song, Z., Zeng, Z., Sun, J., & Song, Y. (2025). Creating a Novel Attention-Enhanced Framework for Video-Based Action Quality Assessment. Sci, 7(2), 54. https://doi.org/10.3390/sci7020054

Article Menu

Creating a Novel Attention-Enhanced Framework for Video-Based Action Quality Assessment

Abstract

Share and Cite

Article Metrics

Article Access Statistics

Further Information

Guidelines

MDPI Initiatives

Follow MDPI