CIGMO: Categorical invariant representations in a deep generative framework

TL;DR

CIGMO is a deep generative model that learns to disentangle object category, shape, and view factors from images, improving shape categorization and downstream tasks like one-shot identification.

Contribution

The paper introduces CIGMO, a novel model that jointly learns category, shape, and view representations with weak supervision, surpassing previous methods in invariant shape clustering.

Findings

Effectively discovers object categories despite view variation.

Outperforms previous invariant clustering algorithms.

Enhances shape representation for downstream tasks.

Abstract

Data of general object images have two most common structures: (1) each object of a given shape can be rendered in multiple different views, and (2) shapes of objects can be categorized in such a way that the diversity of shapes is much larger across categories than within a category. Existing deep generative models can typically capture either structure, but not both. In this work, we introduce a novel deep generative model, called CIGMO, that can learn to represent category, shape, and view factors from image data. The model is comprised of multiple modules of shape representations that are each specialized to a particular category and disentangled from view representation, and can be learned using a group-based weakly supervised learning method. By empirical investigation, we show that our model can effectively discover categories of object shapes despite large view variation and quantitatively supersede various previous methods including the state-of-the-art invariant clustering algorithm. Further, we show that our approach using category-specialization can enhance the learned shape representation to better perform down-stream tasks such as one-shot object identification as well as shape-view disentanglement.