Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces SegWave, a hybrid framework that enhances image tampering detection by integrating a transformer-based architecture with the Discrete Wavelet Transform (DWT) to analyze both spatial and frequency-domain cues. The method employs an Adaptive Sub-band Attention module (ASA) to prioritize high-frequency wavelet components, improving the localization of tampered regions. Experimental results show that SegWave significantly outperforms existing state-of-the-art methods across various challenging benchmarks, underscoring its effectiveness in real-world applications.
SegWave outperforms traditional tampering detection methods by leveraging both spatial and frequency-domain insights, revealing localized manipulations that others miss.
Verifying image authenticity is increasingly difficult, posing serious risks across journalism, law enforcement, and political domains. Most existing forensic methods rely on high-level visual artifacts and treat frame detection as a simple binary task. To address this, we propose SegWave, a hybrid framework that jointly leverages spatial and frequency-domain cues for image tampering detection. SegWave integrates a transformer-based architecture with the Discrete Wavelet Transform (DWT) to capture localized, multi-scale frequency inconsistencies indicative of manipulation. To further improve localization effectiveness, we introduce an Adaptive Sub-band Attention module (ASA) that dynamically highlights the informative high-frequency wavelet components. Extensive experiments on multiple benchmark datasets demonstrate that SegWave consistently outperforms state-of-the-art tampering detection methods in challenging evaluation settings.