Vision Forgery Trace Enhanced VLMs for Generalized AIGC Video Detection
International Journal of Digital Crime and Forensics · 04.03.2026 · 18 (1) · 1-23
DOI: 10.4018/ijdcf.403419 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Digital Crime and Forensics, erschienen am 04.03.2026. Verfasst von Lihua Wang und 4 weiteren Autor:innen. Die Arbeit nennt 67 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 3.0).
Abstract (Originalsprache)
Large vision language models (VLMs) show strong open-world generalization but degrade at domain-specific tasks, while traditional small forensic models perform well on in-distribution datasets yet lack cross-distribution generalization and language-based interpretability. To address this gap, the authors propose a vision forgery trace (VFT)-VLM framework, which incorporates forensic features into a VLM without sacrificing its general reasoning ability. Specifically, a lightweight VFT extraction module learns to encode texture anomalies, edge incoherence, pixel artifacts, and frequency-domain deviations. The traces are incorporated into the InternVL2-8B backbone via low rank adaptation fine-tuning, achieving alignment between visual evidence and textual explanations. Across 14 diverse artificial intelligence-generated content benchmark datasets, VFT-VLM outperforms VLM-based large-scale models and achieves comparable or superior performance to relevant traditional small-scale models. Ablation studies confirm both VFT extraction and low rank adaptation fine-tuning are critical to the performance gains.
Zitieren
APA
Wang, L., Pei, P., He, Y., Huang, Z., & Hu, S. (2026). Vision Forgery Trace Enhanced VLMs for Generalized AIGC Video Detection. International Journal of Digital Crime and Forensics, 18(1), 1-23. https://doi.org/10.4018/ijdcf.403419
Vancouver
Wang L, Pei P, He Y, Huang Z, Hu S. Vision Forgery Trace Enhanced VLMs for Generalized AIGC Video Detection. International Journal of Digital Crime and Forensics. 2026;18(1):1-23. doi:10.4018/ijdcf.403419
BibTeX
@article{Wang2026,
author = {Wang, Lihua and Pei, Pengfei and He, Yiran and Huang, Zihuan and Hu, Shuai},
title = {Vision Forgery Trace Enhanced VLMs for Generalized AIGC Video Detection},
journal = {International Journal of Digital Crime and Forensics},
year = {2026},
volume = {18},
number = {1},
pages = {1--23},
doi = {10.4018/ijdcf.403419}
}
EUVIMED