LatentCLR: A Contrastive Learning Approach for Unsupervised Discovery of Interpretable Directions

TL;DR

This paper introduces LatentCLR, a contrastive learning method that unsupervisedly discovers interpretable directions in GAN latent spaces, enabling semantic image editing without manual annotations.

Contribution

It presents a novel self-supervised contrastive learning approach for discovering semantic directions in GAN latent spaces, eliminating the need for manual labels.

Findings

Achieves comparable semantic directions to state-of-the-art methods.

Enables controllable image editing without supervision.

Demonstrates effectiveness across multiple GAN models.

Abstract

Recent research has shown that it is possible to find interpretable directions in the latent spaces of pre-trained Generative Adversarial Networks (GANs). These directions enable controllable image generation and support a wide range of semantic editing operations, such as zoom or rotation. The discovery of such directions is often done in a supervised or semi-supervised manner and requires manual annotations which limits their use in practice. In comparison, unsupervised discovery allows finding subtle directions that are difficult to detect a priori. In this work, we propose a contrastive learning-based approach to discover semantic directions in the latent space of pre-trained GANs in a self-supervised manner. Our approach finds semantically meaningful dimensions comparable with state-of-the-art methods.