Context-Aware Full Body Anonymization using Text-to-Image Diffusion Models

TL;DR

This paper introduces a method for full body person anonymization using text-to-image diffusion models like Stable Diffusion, achieving high-quality, detailed anonymized images that outperform existing methods in various metrics.

Contribution

The paper presents a novel workflow utilizing diffusion models for full body anonymization, demonstrating superior image quality and invariance across different generative models.

Findings

Outperforms state-of-the-art anonymization methods in image quality, resolution, IS, and FID.

Compatible with multiple diffusion models, ensuring broad applicability.

Produces highly detailed anonymized images that preserve essential features.

Abstract

Anonymization plays a key role in protecting sensible information of individuals in real world datasets. Self-driving cars for example need high resolution facial features to track people and their viewing direction to predict future behaviour and react accordingly. In order to protect people's privacy whilst keeping important features in the dataset, it is important to replace the full body of a person with a highly detailed anonymized one. In contrast to doing face anonymization, full body replacement decreases the ability of recognizing people by their hairstyle or clothes. In this paper, we propose a workflow for full body person anonymization utilizing Stable Diffusion as a generative backend. Text-to-image diffusion models, like Stable Diffusion, OpenAI's DALL-E or Midjourney, have become very popular in recent time, being able to create photorealistic images from a single text prompt. We show that our method outperforms state-of-the art anonymization pipelines with respect to image quality, resolution, Inception Score (IS) and Frechet Inception Distance (FID). Additionally, our method is invariant with respect to the image generator and thus able to be used with the latest models available.