Which Face and Whose Identity? Solving the Dual Challenge of Deepfake Proactive Forensics in Multi-Face Scenarios

2026-04-29

AI Summary

This work addresses the challenge of simultaneously localizing forged faces and tracing their source identities in multi-person scenarios—a task where existing deepfake active forensic methods fall short. To this end, we propose the Deeply Attributable Watermarking Framework (DAWF), which leverages a multi-face encoding-decoding architecture to enable parallel watermark embedding and cross-face collaborative processing. A key innovation is the introduction of a selective region-aware supervision loss that guides the decoder to focus explicitly on manipulated regions. Operating in an end-to-end manner without offline preprocessing, DAWF jointly infers both “which face has been forged” and “whose identity was impersonated.” Experiments demonstrate that DAWF significantly improves accuracy in forgery localization and identity attribution on complex multi-face datasets, offering robust defense against deepfakes in real-world multi-person interaction settings.

Abstract

Unlike single-face forgeries, deepfakes in complex multi-person interaction scenarios (such as group photos and multi-person meetings) more closely reflect real-world threats. Although existing proactive forensics solutions demonstrate good performance, they heavily rely on a "single-face" setting, making it difficult to effectively address the problems of deepfake localization and source tracing in complex multi-person environments. To address this challenge, we propose the Deep Attributable Watermarking Framework (DAWF). This framework adopts a novel multi-face encoder-decoder architecture that bypasses the cumbersome offline pre-processing steps of traditional forensics, facilitating efficient in-network parallel watermark embedding and cross-face collaborative processing. Crucially, we propose a selective regional supervision loss. This innovative mechanism guides the decoder to focus exclusively on the facial regions tampered with by deepfakes. Leveraging this mechanism alongside the embedded identity payloads, DAWF realizes the "which + who" goal, answering the dual questions of which facial region was forged and who was forged. Extensive experiments on challenging multi-face datasets show that DAWF achieves excellent deepfake localization and traceability in complex multi-person scenes.

Problem

• deepfake
• multi-face
• proactive forensics
• localization
• source tracing

Innovation

• multi-face deepfake forensics
• attributable watermarking
• selective regional supervision
• identity traceability
• in-network watermark embedding