Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FlowID : Enhancing Forensic Identification with Latent Flow-Matching Models

Jules Ripoll, David Bertoin, Alasdair Newson, Charles Dossal, Jose Pablo Baraybar

Abstract

Every day, many people die under violent circumstances, whether from crimes, war, migration, or climate disasters. Medico-legal and law enforcement institutions document many portraits of the deceased for evidence, but cannot immediately carry out identification on them. While traditional image editing tools can process these photos for public release, the workflow is lengthy and produces suboptimal results. In this work, we leverage advances in image generation models, which can now produce photorealistic human portraits, to introduce FlowID, an identity-preserving facial reconstruction method. Our approach combines single-image fine-tuning, which adapts the generative model to out-of-distribution injured faces, with attention-based masking that localizes edits to damaged regions while preserving identity-critical features. Together, these components enable the removal of artifacts from violent death while retaining sufficient identity information to support identification. To evaluate our method, we introduce InjuredFaces, a novel benchmark for identity-preserving facial reconstruction under severe facial damage. Beyond serving as an evaluation tool for this work, InjuredFaces provides a standardized resource for the community to study and compare methods addressing facial reconstruction in extreme conditions. Experimental results show that FlowID outperforms state-of-the-art open-source methods while maintaining low memory requirements, making it suitable for local deployment without compromising data privacy.

FlowID : Enhancing Forensic Identification with Latent Flow-Matching Models

Abstract

Every day, many people die under violent circumstances, whether from crimes, war, migration, or climate disasters. Medico-legal and law enforcement institutions document many portraits of the deceased for evidence, but cannot immediately carry out identification on them. While traditional image editing tools can process these photos for public release, the workflow is lengthy and produces suboptimal results. In this work, we leverage advances in image generation models, which can now produce photorealistic human portraits, to introduce FlowID, an identity-preserving facial reconstruction method. Our approach combines single-image fine-tuning, which adapts the generative model to out-of-distribution injured faces, with attention-based masking that localizes edits to damaged regions while preserving identity-critical features. Together, these components enable the removal of artifacts from violent death while retaining sufficient identity information to support identification. To evaluate our method, we introduce InjuredFaces, a novel benchmark for identity-preserving facial reconstruction under severe facial damage. Beyond serving as an evaluation tool for this work, InjuredFaces provides a standardized resource for the community to study and compare methods addressing facial reconstruction in extreme conditions. Experimental results show that FlowID outperforms state-of-the-art open-source methods while maintaining low memory requirements, making it suitable for local deployment without compromising data privacy.

Paper Structure

This paper contains 19 sections, 6 equations, 9 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Training-free methods.
  4. Training-based methods.
  5. Method
  6. Background
  7. Latent Flow-Matching Models.
  8. Image Inversion.
  9. Single-Image Fine-tuning
  10. Localized Editing with Inferred Masks
  11. Experiments
  12. The InjuredFaces Benchmark
  13. Results
  14. Additional Results on FFHQ
  15. Ablation study.
  16. ...and 4 more sections

Figures (9)

  • Figure 1: Samples from InjuredFaces, our facial reconstruction benchmark, edited with FlowID. Top: original images. Bottom: edited results.
  • Figure 2: Single-image fine-tuning improves inversion performance.
  • Figure 3: Impact of our masking component for details preservation. Without masking, the person's tattoo is altered (middle) When using our masking component, we manage to infer the location of the tattoo and explicitly preserve it (right).
  • Figure 4: Qualitative comparison of image editing methods on InjuredFaces.
  • Figure 5: Edit performance as a function of fine-tuning steps.
  • ...and 4 more figures