Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

My Body My Choice: Human-Centric Full-Body Anonymization

Umur Aybars Ciftci, Ali Kemal Tanriverdi, Ilke Demir

TL;DR

''My Body My Choice'' (MBMC) enables physical and adversarial anonymization by removal and swapping approaches aimed for four tasks, designed by single or multi, ControlNet or GAN modules, combining several diffusion models.

Abstract

In an era of increasing privacy concerns for our online presence, we propose that the decision to appear in a piece of content should only belong to the owner of the body. Although some automatic approaches for full-body anonymization have been proposed, human-guided anonymization can adapt to various contexts, such as cultural norms, personal relations, esthetic concerns, and security issues. ''My Body My Choice'' (MBMC) enables physical and adversarial anonymization by removal and swapping approaches aimed for four tasks, designed by single or multi, ControlNet or GAN modules, combining several diffusion models. We evaluate anonymization on seven datasets; compare with SOTA inpainting and anonymization methods; evaluate by image, adversarial, and generative metrics; and conduct reidentification experiments.

My Body My Choice: Human-Centric Full-Body Anonymization

TL;DR

''My Body My Choice'' (MBMC) enables physical and adversarial anonymization by removal and swapping approaches aimed for four tasks, designed by single or multi, ControlNet or GAN modules, combining several diffusion models.

Abstract

In an era of increasing privacy concerns for our online presence, we propose that the decision to appear in a piece of content should only belong to the owner of the body. Although some automatic approaches for full-body anonymization have been proposed, human-guided anonymization can adapt to various contexts, such as cultural norms, personal relations, esthetic concerns, and security issues. ''My Body My Choice'' (MBMC) enables physical and adversarial anonymization by removal and swapping approaches aimed for four tasks, designed by single or multi, ControlNet or GAN modules, combining several diffusion models. We evaluate anonymization on seven datasets; compare with SOTA inpainting and anonymization methods; evaluate by image, adversarial, and generative metrics; and conduct reidentification experiments.
Paper Structure (15 sections, 5 figures, 3 tables)

This paper contains 15 sections, 5 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Relevant Work
  3. My Body My Choice
  4. System Overview
  5. Physical Removal
  6. Adversarial Removal
  7. Mask-based Removal
  8. Identity Removal
  9. Multi-body Anonymization
  10. Results
  11. Physical Removal
  12. Adversarial Removal
  13. Mask-based Removal
  14. Identity Removal
  15. Conclusion

Figures (5)

  • Figure 2: Mask-based Removal Pipeline. (Left) Embedding search, preparation of multiControlNet input, and face enhancements. (Right) Running the system independent of the body order for a multi-body image.
  • Figure 3: Removal Comparison. Compared to MAT mat and LaMa lama, MBMC creates less visual artifacts for patterned (top), extreme-pose (middle), and complex scene (bottom) cases.
  • Figure 4: Mask-based Removal Comparison. Original, deepprivacy2, and ours are compared visually on MPII samples. Our results are more detailed, structured, contextual, and less similar to original.
  • Figure 5: Embedding Ablation. MBMC currently uses the bottom mid, significantly different appearance with a similar pose.
  • Figure 6: Surveillance Anonymization. MBMC works on any domain, size, resolution images, decreasing reidentification.