HYDRO: Towards Non-Reversible Face De-Identification Using a High-Fidelity Hybrid Diffusion and Target-Oriented Approach

Researchers introduced HYDRO, a target-oriented face de-identification method that combines generative de-identification with a dedicated diffusion process to prevent reconstruction attacks, according to an arXiv paper (arXiv:2609.27011v1). HYDRO de-identifies a face image, injects noise to impede reconstruction, then applies diffusion-based recovery to restore fidelity, and adds a novel Eye Similarity Discriminator (ESD) to training to retain gaze direction. Across three datasets, HYDRO reduced reconstruction attack success by 85.7% on average versus multiple state-of-the-art competitors while matching their fidelity and attribute retention.

arXiv:2609.27011v1 Announce Type: new Abstract: Target-oriented face de-identification models aim to anonymize the identity of a target individual across different images or video frames, such that the target can no longer be reliably recognized, while maintaining key characteristics of the visual data. Such models commonly leverage generative encoder-decoder architectures to manipulate facial appearances, enabling them to produce realistic high-fidelity de-identification results, while ensuring considerable attribute-retention capabilities. However, target-oriented models also carry the risk of inadvertently preserving subtle identity cues, making them potentially reversible and susceptible to reconstruction attacks. To address this problem, we introduce in this paper a novel robust face de-identification approach, called HYDRO, that combines target-oriented models with a dedicated diffusion process specifically designed to destroy any imperceptible information that may allow learning to reverse the de-identification procedure. HYDRO first de-identifies the given face image, injects noise into the de-identification result to impede reconstruction, and then applies a diffusion-based recovery step to improve fidelity and minimize the impact of the noising process on the data characteristics. To further improve image fidelity and better retain gaze directions, a novel Eye Similarity Discriminator ESD is also introduced and incorporated it into the training of HYDRO. Extensive quantitative and qualitative experiments on three diverse datasets demonstrate that HYDRO exhibits state-of-the-art SOTA fidelity and attribute-retention capabilities, while being the only target-oriented method resilient against reconstruction attacks. In comparison to multiple SOTA competitors, HYDRO reduces the success of reconstruction attacks by 85.7% on average.