Discovering Multiple and Diverse Directions for Cognitive Image Properties

TL;DR

This paper introduces a framework for discovering multiple, diverse directions in GAN latent spaces to manipulate cognitive image properties like memorability and aesthetics, enabling more controllable and varied image editing.

Contribution

It proposes a novel method for identifying multiple diverse directions for specific cognitive properties in GANs, enhancing interpretability and control over image generation.

Findings

Successfully manipulates memorability, emotional valence, and aesthetics.

Produces diverse outputs for each property.

Demonstrates effectiveness through extensive experiments.

Abstract

Recent research has shown that it is possible to find interpretable directions in the latent spaces of pre-trained GANs. These directions enable controllable generation and support a variety of semantic editing operations. While previous work has focused on discovering a single direction that performs a desired editing operation such as zoom-in, limited work has been done on the discovery of multiple and diverse directions that can achieve the desired edit. In this work, we propose a novel framework that discovers multiple and diverse directions for a given property of interest. In particular, we focus on the manipulation of cognitive properties such as Memorability, Emotional Valence and Aesthetics. We show with extensive experiments that our method successfully manipulates these properties while producing diverse outputs. Our project page and source code can be found at http://catlab-team.github.io/latentcognitive.