Denoise and Align: Diffusion-Driven Foreground Knowledge Prompting for Open-Vocabulary Temporal Action Detection

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhu, Sa, Zhang, Wanqian, Wang, Lin, Zhang, Jinchao, Wang, Cong, Li, Bo
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866908980340588544
author Zhu, Sa
Zhang, Wanqian
Wang, Lin
Zhang, Jinchao
Wang, Cong
Li, Bo
author_facet Zhu, Sa
Zhang, Wanqian
Wang, Lin
Zhang, Jinchao
Wang, Cong
Li, Bo
contents Open-Vocabulary Temporal Action Detection (OV-TAD) aims to localize and classify action segments of unseen categories in untrimmed videos, where effective alignment between action semantics and video representations is critical for accurate detection. However, existing methods struggle to mitigate the semantic imbalance between concise, abstract action labels and rich, complex video contents, inevitably introducing semantic noise and misleading cross-modal alignment. To address this challenge, we propose DFAlign, the first framework that leverages diffusion-based denoising to generate foreground knowledge for the guidance of action-video alignment. Following the 'conditioning, denoising and aligning' manner, we first introduce the Semantic-Unify Conditioning (SUC) module, which unifies action-shared and action-specific semantics as conditions for diffusion denoising. Then, the Background-Suppress Denoising (BSD) module generates foreground knowledge by progressively removing background redundancy from videos through denoising process. This foreground knowledge serves as effective intermediate semantic anchor between video and text representations, mitigating the semantic gap and enhancing the discriminability of action-relevant segments. Furthermore, we introduce the Foreground-Prompt Alignment (FPA) module to inject extracted foreground knowledge as prompt tokens into text representations, guiding model's attention towards action-relevant segments and enabling precise cross-modal alignment. Extensive experiments demonstrate that our method achieves state-of-the-art performance on two OV-TAD benchmarks. The code repository is provided as follows: https://anonymous.4open.science/r/Code-2114/.
format Preprint
id arxiv_https___arxiv_org_abs_2604_18313
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Denoise and Align: Diffusion-Driven Foreground Knowledge Prompting for Open-Vocabulary Temporal Action Detection
Zhu, Sa
Zhang, Wanqian
Wang, Lin
Zhang, Jinchao
Wang, Cong
Li, Bo
Computer Vision and Pattern Recognition
Open-Vocabulary Temporal Action Detection (OV-TAD) aims to localize and classify action segments of unseen categories in untrimmed videos, where effective alignment between action semantics and video representations is critical for accurate detection. However, existing methods struggle to mitigate the semantic imbalance between concise, abstract action labels and rich, complex video contents, inevitably introducing semantic noise and misleading cross-modal alignment. To address this challenge, we propose DFAlign, the first framework that leverages diffusion-based denoising to generate foreground knowledge for the guidance of action-video alignment. Following the 'conditioning, denoising and aligning' manner, we first introduce the Semantic-Unify Conditioning (SUC) module, which unifies action-shared and action-specific semantics as conditions for diffusion denoising. Then, the Background-Suppress Denoising (BSD) module generates foreground knowledge by progressively removing background redundancy from videos through denoising process. This foreground knowledge serves as effective intermediate semantic anchor between video and text representations, mitigating the semantic gap and enhancing the discriminability of action-relevant segments. Furthermore, we introduce the Foreground-Prompt Alignment (FPA) module to inject extracted foreground knowledge as prompt tokens into text representations, guiding model's attention towards action-relevant segments and enabling precise cross-modal alignment. Extensive experiments demonstrate that our method achieves state-of-the-art performance on two OV-TAD benchmarks. The code repository is provided as follows: https://anonymous.4open.science/r/Code-2114/.
title Denoise and Align: Diffusion-Driven Foreground Knowledge Prompting for Open-Vocabulary Temporal Action Detection
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.18313