Submit

Login Login

Register Register

Home / Journals / CMC / Online First / doi:10.32604/cmc.2025.072964
Journal Menu
Special Issues
Table of Content

All issues

Online First

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Open Access

ARTICLE

DyLoRA-TAD: Dynamic Low-Rank Adapter for End-to-End Temporal Action Detection

Jixin Wu1,2, Mingtao Zhou2,3, Di Wu2,3, Wenqi Ren4, Jiatian Mei2,3, Shu Zhang1,*
1 School of Information Science and Technology, Yunnan Normal University, Kunming, 650500, China
2 Yunnan Key Laboratory of Smart Education, Yunnan Normal University, Kunming, 650500, China
3 Key Laboratory of Education Informatization for Nationalities, Ministry of Education, Yunnan Normal University, Kunming, 650500, China
4 School of Cyber Science and Technology, Sun Yat-sen University, Guangzhou, 510275, China
* Corresponding Author: Shu Zhang. Email: email
(This article belongs to the Special Issue: Advances in Action Recognition: Algorithms, Applications, and Emerging Trends)

Computers, Materials & Continua https://doi.org/10.32604/cmc.2025.072964

Received 08 September 2025; Accepted 12 November 2025; Published online 08 December 2025

Abstract

End-to-end Temporal Action Detection (TAD) has achieved remarkable progress in recent years, driven by innovations in model architectures and the emergence of Video Foundation Models (VFMs). However, existing TAD methods that perform full fine-tuning of pretrained video models often incur substantial computational costs, which become particularly pronounced when processing long video sequences. Moreover, the need for precise temporal boundary annotations makes data labeling extremely expensive. In low-resource settings where annotated samples are scarce, direct fine-tuning tends to cause overfitting. To address these challenges, we introduce Dynamic Low-Rank Adapter (DyLoRA), a lightweight fine-tuning framework tailored specifically for the TAD task. Built upon the Low-Rank Adaptation (LoRA) architecture, DyLoRA adapts only the key layers of the pretrained model via low-rank decomposition, reducing the number of trainable parameters to less than 5% of full fine-tuning methods. This significantly lowers memory consumption and mitigates overfitting in low-resource settings. Notably, DyLoRA enhances the temporal modeling capability of pretrained models by optimizing temporal dimension weights, thereby alleviating the representation misalignment of temporal features. Experimental results demonstrate that DyLoRA-TAD achieves impressive performance, with 73.9% mAP on THUMOS14, 39.52% on ActivityNet-1.3, and 28.2% on Charades, substantially surpassing the best traditional feature-based methods.

Keywords

Temporal action detection; end-to-end training; dynamic low-rank adapter; parameter-efficient fine-tuning; video understanding

  • 86

    View

  • 10

    Download

  • 1

    Like

Related articles
Copyright© 2025 Tech Science Press
© 1997-2025 TSP (Henderson, USA) unless otherwise stated

Share Link