Visual Concepts Tokenization

TL;DR

This paper introduces VCT, an unsupervised transformer framework that tokenizes images into disentangled visual concepts, improving scene understanding and representation learning.

Contribution

VCT is the first unsupervised transformer-based method to produce independent visual concept tokens using cross-attention and a novel disentangling loss.

Findings

VCT achieves state-of-the-art results on multiple datasets.

VCT effectively disentangles visual concepts without supervision.

VCT improves scene decomposition accuracy.

Abstract

Obtaining the human-like perception ability of abstracting visual concepts from concrete pixels has always been a fundamental and important target in machine learning research fields such as disentangled representation learning and scene decomposition. Towards this goal, we propose an unsupervised transformer-based Visual Concepts Tokenization framework, dubbed VCT, to perceive an image into a set of disentangled visual concept tokens, with each concept token responding to one type of independent visual concept. Particularly, to obtain these concept tokens, we only use cross-attention to extract visual information from the image tokens layer by layer without self-attention between concept tokens, preventing information leakage across concept tokens. We further propose a Concept Disentangling Loss to facilitate that different concept tokens represent independent visual concepts. The cross-attention and disentangling loss play the role of induction and mutual exclusion for the concept tokens, respectively. Extensive experiments on several popular datasets verify the effectiveness of VCT on the tasks of disentangled representation learning and scene decomposition. VCT achieves the state of the art results by a large margin.