Identity-preserving Editing of Multiple Facial Attributes by Learning Global Edit Directions and Local Adjustments

TL;DR

This paper introduces ID-Style, a novel GAN-based architecture that effectively preserves identity during facial attribute editing by learning global directions and local adjustments, outperforming existing methods in identity preservation and attribute manipulation accuracy.

Contribution

ID-Style combines learnable global directions and instance-aware local adjustments to improve identity preservation in facial attribute editing, with a significantly smaller network size.

Findings

Outperforms baselines by 10% in identity preservation (FRS)

Achieves 7% higher average accuracy in attribute manipulation

Reduces network size by approximately 95%

Abstract

Semantic facial attribute editing using pre-trained Generative Adversarial Networks (GANs) has attracted a great deal of attention and effort from researchers in recent years. Due to the high quality of face images generated by StyleGANs, much work has focused on the StyleGANs' latent space and the proposed methods for facial image editing. Although these methods have achieved satisfying results for manipulating user-intended attributes, they have not fulfilled the goal of preserving the identity, which is an important challenge. We present ID-Style, a new architecture capable of addressing the problem of identity loss during attribute manipulation. The key components of ID-Style include Learnable Global Direction (LGD), which finds a shared and semi-sparse direction for each attribute, and an Instance-Aware Intensity Predictor (IAIP) network, which finetunes the global direction according to the input instance. Furthermore, we introduce two losses during training to enforce the LGD to find semi-sparse semantic directions, which along with the IAIP, preserve the identity of the input instance. Despite reducing the size of the network by roughly 95% as compared to similar state-of-the-art works, it outperforms baselines by 10% and 7% in Identity preserving metric (FRS) and average accuracy of manipulation (mACC), respectively.