# The Devil Is in the Leakage: A Disentangled Dual-Purification Framework for High-Fidelity Hairstyle Transfer

arXiv: https://arxiv.org/abs/2607.11281 · HTML: https://mlsanity.com/paper/2607.11281/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 30 of its 53 references that are in the map.
- First posted on arXiv 13 Jul 2026.

## Builds on (most cited first, 30 of 30)

- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [The Unreasonable Effectiveness of Deep Features as a Perceptual Metric](https://mlsanity.com/paper/1801.03924/) (2018)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [A Style-Based Generator Architecture for Generative Adversarial Networks](https://mlsanity.com/paper/1812.04948/) (2018)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Analyzing and Improving the Image Quality of StyleGAN](https://mlsanity.com/paper/1912.04958/) (2019)
- [Progressive Growing of GANs for Improved Quality, Stability, and Variation](https://mlsanity.com/paper/1710.10196/) (2017)
- [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://mlsanity.com/paper/2305.06500/) (2023)
- [InstructPix2Pix: Learning to Follow Image Editing Instructions](https://mlsanity.com/paper/2211.09800/) (2022)
- [Shortcut Learning in Deep Neural Networks](https://mlsanity.com/paper/2004.07780/) (2020)
- [Unsupervised Domain Adaptation by Backpropagation](https://mlsanity.com/paper/1409.7495/) (2014)
- [Alias-Free Generative Adversarial Networks](https://mlsanity.com/paper/2106.12423/) (2021)
- [Adversarial Examples Are Not Bugs, They Are Features](https://mlsanity.com/paper/1905.02175/) (2019)
- [MINE: Mutual Information Neural Estimation](https://mlsanity.com/paper/1801.04062/) (2018)
- [BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation](https://mlsanity.com/paper/1808.00897/) (2018)
- [Disentangling by Factorising](https://mlsanity.com/paper/1802.05983/) (2018)
- [Diverse Image-to-Image Translation via Disentangled Representations](https://mlsanity.com/paper/1808.00948/) (2018)
- [Disentangled and Controllable Face Image Generation via 3D Imitative-Contrastive Learning](https://mlsanity.com/paper/2004.11660/) (2020)
- [HairCLIP: Design Your Hair by Text and Reference Image](https://mlsanity.com/paper/2112.05142/) (2021)
- [Semantic Human Matting](https://mlsanity.com/paper/1809.01354/) (2018)
- [A Unified Feature Disentangler for Multi-Domain Image Translation and Manipulation](https://mlsanity.com/paper/1809.01361/) (2018)
- [HairCLIPv2: Unifying Hair Editing via Proxy Feature Blending](https://mlsanity.com/paper/2310.10651/) (2023)
- [Style Your Hair: Latent Optimization for Pose-Invariant Hairstyle Transfer via Local-Style-Aware Hair Alignment](https://mlsanity.com/paper/2208.07765/) (2022)
- [3D-aware Blending with Generative NeRFs](https://mlsanity.com/paper/2302.06608/) (2023)
- [DIFFER: Disentangling Identity Features via Semantic Cues for Clothes-Changing Person Re-ID](https://mlsanity.com/paper/2503.22912/) (2025)
- [Reinforced Disentanglement for Face Swapping without Skip Connection](https://mlsanity.com/paper/2307.07928/) (2023)
- [What to Preserve and What to Transfer: Faithful, Identity-Preserving Diffusion-based Hairstyle Transfer](https://mlsanity.com/paper/2408.16450/) (2024)
- [AFD-Net: Adaptive Fully-Dual Network for Few-Shot Object Detection](https://mlsanity.com/paper/2011.14667/) (2020)
- [Stable-Hair v2: Real-World Hair Transfer via Multiple-View Diffusion Model](https://mlsanity.com/paper/2507.07591/) (2025)

## Built on by (most cited first, 2 of 2)

- [Knowing You at First Glance: Inferring Apparent Personality from Faces](https://mlsanity.com/paper/2607.14631/) (2026)
- [Talk Like You: Imitating How You Speak in Real-Time Talking Head Generation](https://mlsanity.com/paper/2610.06658/) (2026)

## Abstract (arXiv)

Hairstyle transfer aims to synthesize a photorealistic portrait by transplanting the hairstyle from a reference image onto a source subject while preserving the source identity. Recent foundation models show strong generative capability, but they struggle with the zero-shot disentanglement required for precise local editing, often entangling the reference hairstyle with its original identity and pose. Existing diffusion-based pipelines typically decompose the task by first generating a "bald" image from the source and then injecting hairstyle features from the reference. However, we show that this paradigm suffers from a fundamental leakage problem. Identity Leakage in Hairstyle occurs when hairstyle features retain reference identity or pose information, while Flaw Leakage in Bald arises when residual artifacts in the bald image are propagated into the final synthesis. To address both issues, we propose the Dual-Purification Framework (DPF), which introduces two complementary training-time regularizers. Adversarial Hairstyle Purification (AHP) purifies hairstyle features by suppressing identity predictability under a mutual-information-inspired adversarial objective. Contrastive Geometric Purification (CGP) regularizes the ControlNet pathway with a contrastive objective, reducing the model's reliance on geometric artifacts in the bald condition. By jointly purifying the hairstyle representation and geometric pathway, DPF achieves high-fidelity, identity-preserving hairstyle transfer and state-of-the-art performance on diverse benchmarks.