🤖 AI Summary
To address the privacy-compliance dilemma in facial data collection—where individual consent is mandatory yet high-quality annotated data remains difficult to obtain—this paper proposes a three-stage face anonymization framework based on diffusion models. Methodologically, it introduces BrushNet, the first text-guided regeneration model that jointly incorporates semantic textual prompts (e.g., age, gender, expression) and facial attribute control, enabling image+mask+text co-conditioning without GAN training. Integrated with RetinaNet for precise facial region detection and VGG-Face for perceptual feature evaluation, the framework achieves fine-grained, controllable synthesis. The key contribution is the first systematic application of text-driven diffusion models to face anonymization, achieving strong de-identification guarantees while significantly improving visual naturalness and utility for downstream vision tasks (e.g., detection, recognition), thereby reconciling stringent privacy protection with high-fidelity reconstruction.
📝 Abstract
The increasing prevalence of computer vision applications necessitates handling vast amounts of visual data, often containing personal information. While this technology offers significant benefits, it should not compromise privacy. Data privacy regulations emphasize the need for individual consent for processing personal data, hindering researchers' ability to collect high-quality datasets containing the faces of the individuals. This paper presents a deep learning-based face anonymization pipeline to overcome this challenge. Unlike most of the existing methods, our method leverages recent advancements in diffusion-based inpainting models, eliminating the need for training Generative Adversarial Networks. The pipeline employs a three-stage approach: face detection with RetinaNet, feature extraction with VGG-Face, and realistic face generation using the state-of-the-art BrushNet diffusion model. BrushNet utilizes the entire image, face masks, and text prompts specifying desired facial attributes like age, ethnicity, gender, and expression. This enables the generation of natural-looking images with unrecognizable individuals, facilitating the creation of privacy-compliant datasets for computer vision research.