Search papers, labs, and topics across Lattice.
MD-Face, a novel GAN-based framework, achieves label-free disentangled representation learning for facial attribute editing by employing a Mixture of Experts (MoE) backbone with dynamic expert allocation. A geometry-aware loss function, aligning semantic vectors with Semantic Boundary Vectors (SBV) via a Jacobian-based pushforward, further reduces attribute entanglement. Experiments on ProGAN and StyleGAN demonstrate that MD-Face achieves comparable performance to supervised methods while surpassing unsupervised baselines, offering improved image quality and lower latency than diffusion-based approaches.
GANs can now edit faces with disentangled attributes *without* needing labeled data, rivaling supervised methods in quality and speed.
GAN-based facial attribute editing is widely used in virtual avatars and social media but often suffers from attribute entanglement, where modifying one face attribute unintentionally alters others. While supervised disentangled representation learning can address this, it relies heavily on labeled data, incurring high annotation costs. To address these challenges, we propose MD-Face, a label-free disentangled representation learning framework based on Mixture of Experts (MoE). MD-Face utilizes a MoE backbone with a gating mechanism that dynamically allocates experts, enabling the model to learn semantic vectors with greater independence. To further enhance attribute entanglement, we introduce a geometry-aware loss, which aligns each semantic vector with its corresponding Semantic Boundary Vector (SBV) through a Jacobian-based pushforward method. Experiments with ProGAN and StyleGAN show that MD-Face outperforms unsupervised baselines and competes with supervised ones. Compared to diffusion-based methods, it offers better image quality and lower inference latency, making it ideal for interactive editing.
