DisCont: Self-Supervised Visual Attribute Disentanglement using Context Vectors

TL;DR

DisCont is a self-supervised framework that leverages contrastive learning and structural biases to disentangle multiple image attributes without supervision, enhancing interpretability and control.

Contribution

We introduce DisCont, a novel self-supervised method that combines contrastive learning with structural biases for attribute disentanglement in images.

Findings

Effective on four benchmark datasets

Achieves superior disentanglement quality

Bridges contrastive learning with unsupervised disentanglement

Abstract

Disentangling the underlying feature attributes within an image with no prior supervision is a challenging task. Models that can disentangle attributes well provide greater interpretability and control. In this paper, we propose a self-supervised framework DisCont to disentangle multiple attributes by exploiting the structural inductive biases within images. Motivated by the recent surge in contrastive learning paradigms, our model bridges the gap between self-supervised contrastive learning algorithms and unsupervised disentanglement. We evaluate the efficacy of our approach, both qualitatively and quantitatively, on four benchmark datasets.