Label-guided Facial Retouching Reversion
This addresses the challenge of photo authenticity for social media and forensic applications, but it is incremental as it builds on prior retouching detection and image reversion work.
The paper tackles the problem of reverting facial retouching, including geometric deformations, which existing makeup removal methods cannot handle, by proposing the Re-Face framework with a detector, FaceR model, and H-AdaIN module, achieving effective results as demonstrated in experiments.
With the popularity of social media platforms and retouching tools, more people are beautifying their facial photos, posing challenges for fields requiring photo authenticity. To address this issue, some work has proposed makeup removal methods, but they cannot revert images involving geometric deformations caused by retouching. To tackle the problem of facial retouching reversion, we propose a framework, dubbed Re-Face, which consists of three components: a facial retouching detector, an image reversion model named FaceR, and a color correction module called Hierarchical Adaptive Instance Normalization (H-AdaIN). FaceR can utilize labels generated by the facial retouching detector as guidance to revert the retouched facial images. Then, color correction is performed using H-AdaIN to address the issue of color shift. Extensive experiments demonstrate the effectiveness of our framework and each module.