发表刊物：

Chinese Conference on Pattern Recognition and Computer Vision (PRCV)

关键字：

Image Forensics，Image Tampering Detection，Image Tampering Localization，Diffusion ModelBased Image Tampering

摘要：

Current methods exhibit limited accuracy in detection and localization when facing diffusion model-based image tampering that are semantically coherent yet leave only faint traces. To meet this challenge, we introduce CABFNet, a dualbranch framework for image tampering detection and localization. The imagelevel branch employs tamperingaware adaptation of CLIP model to leverage its powerful global semantic representations for robust tampering detection. The pixellevel branch uses VMamba with MultiForensic Signal Guidance (MFSG) strategy, enabling finegrained and accurate localization of tampering regions. More importantly, we design a CrossAttention Bidirectional Fusion (CABF) module to enable deep interaction and fusion between semantic and pixellevel features across the two branches. Moreover, we construct MAF-set, a large-scale and diverse dataset of diffusion model￾based image tampering, to fill the gap of limited quantity and diversity in existing datasets. Extensive experiments on public benchmarks and on our proposed dataset MAF-set demonstrate that CABF-Net signifi￾cantly outperforms current state-of-the-art approaches in both detection accuracy and localization precision.

论文类型：

CCF C会

期号：

PRCV 2025

页面范围：

328-343

是否译文：

否

发表时间：

2026-01-09

收录刊物：

EI

发布期刊链接：

https://doi.org/10.1007/978-981-95-5702-8_23