DeformTrace: A Deformable State Space Model with Relay Tokens for Temporal Forgery Localization
#DeformTrace #deformable state space #relay tokens #temporal forgery #localization #video manipulation #deepfake detection
📌 Key Takeaways
- DeformTrace is a new model for detecting forged video segments.
- It uses a deformable state space architecture for temporal analysis.
- Relay tokens are introduced to enhance localization of manipulated content.
- The model aims to improve accuracy in identifying fake video sequences.
📖 Full Retelling
arXiv:2603.04882v1 Announce Type: cross
Abstract: Temporal Forgery Localization (TFL) aims to precisely identify manipulated segments in video and audio, offering strong interpretability for security and forensics. While recent State Space Models (SSMs) show promise in precise temporal reasoning, their use in TFL is hindered by ambiguous boundaries, sparse forgeries, and limited long-range modeling. We propose DeformTrace, which enhances SSMs with deformable dynamics and relay mechanisms to add
🏷️ Themes
Video Forensics, AI Detection
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Computer Vision and Pattern Recognition arXiv:2603.04882 [Submitted on 5 Mar 2026] Title: DeformTrace: A Deformable State Space Model with Relay Tokens for Temporal Forgery Localization Authors: Xiaodong Zhu , Suting Wang , Yuanming Zheng , Junqi Yang , Yangxu Liao , Yuhong Yang , Weiping Tu , Zhongyuan Wang View a PDF of the paper titled DeformTrace: A Deformable State Space Model with Relay Tokens for Temporal Forgery Localization, by Xiaodong Zhu and 7 other authors View PDF HTML Abstract: Temporal Forgery Localization aims to precisely identify manipulated segments in video and audio, offering strong interpretability for security and forensics. While recent State Space Models show promise in precise temporal reasoning, their use in TFL is hindered by ambiguous boundaries, sparse forgeries, and limited long-range modeling. We propose DeformTrace, which enhances SSMs with deformable dynamics and relay mechanisms to address these challenges. Specifically, Deformable Self-SSM (DS-SSM) introduces dynamic receptive fields into SSMs for precise temporal localization. To further enhance its capacity for temporal reasoning and mitigate long-range decay, a Relay Token Mechanism is integrated into DS-SSM. Besides, Deformable Cross-SSM (DC-SSM) partitions the global state space into query-specific subspaces, reducing non-forgery information accumulation and boosting sensitivity to sparse forgeries. These components are integrated into a hybrid architecture that combines the global modeling of Transformers with the efficiency of SSMs. Extensive experiments show that DeformTrace achieves state-of-the-art performance with fewer parameters, faster inference, and stronger robustness. Comments: 9 pages, 4 figures, accepted by AAAI 2026 Subjects: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI cs.MM) Cite as: arXiv:2603.04882 [cs.CV] (or arXiv:2603.04882v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2603.04882 Fo...
Read full article at source