diffDemorph: Extending Reference-Free Demorphing to Unseen Faces

TL;DR

diffDeMorph is a diffusion-based method that accurately reverses face morphs into original images, generalizing across morph techniques and face styles, and outperforming previous methods significantly.

Contribution

The paper introduces diffDeMorph, the first diffusion-based RF demorphing approach that generalizes across morph types and face styles, improving accuracy and practicality.

Findings

Outperforms state-of-the-art by ≥59.46% across datasets

Generalizes across morph techniques and face styles

Effective on real morphs trained on synthetic data

Abstract

A face morph is created by combining two face images corresponding to two identities to produce a composite that successfully matches both the constituent identities. Reference-free (RF) demorphing reverses this process using only the morph image, without the need for additional reference images. Previous RF demorphing methods are overly constrained, as they rely on assumptions about the distributions of training and testing morphs such as the morphing technique used (e.g., landmark-based) and face image style (e.g., passport photos). In this paper, we introduce a novel diffusion-based approach, referred to as diffDeMorph, that effectively disentangles component images from a composite morph image with high visual fidelity. Our method is the first to generalize across morph techniques and face styles, beating the current state of the art by $\geq 59.46\%$ under a common training protocol across all datasets tested. We train our method on morphs created using synthetically generated face images and test on real morphs, thereby enhancing the practicality of the technique. Experiments on six datasets and two face matchers establish the utility and efficacy of our method.