Using Shape to Categorize: Low-Shot Learning with an Explicit Shape Bias

TL;DR

This paper explores how explicitly incorporating 3D shape reasoning into low-shot learning enhances generalization, proposing a new shape-based embedding method and introducing the Toys4K dataset for evaluation.

Contribution

It introduces a novel shape-based embedding approach for low-shot learning and presents the Toys4K dataset to support shape reasoning research.

Findings

Improved low-shot learning performance on multiple datasets

Effective use of 3D shape in discriminative embedding space

Introduction of the large-scale Toys4K dataset

Abstract

It is widely accepted that reasoning about object shape is important for object recognition. However, the most powerful object recognition methods today do not explicitly make use of object shape during learning. In this work, motivated by recent developments in low-shot learning, findings in developmental psychology, and the increased use of synthetic data in computer vision research, we investigate how reasoning about 3D shape can be used to improve low-shot learning methods' generalization performance. We propose a new way to improve existing low-shot learning approaches by learning a discriminative embedding space using 3D object shape, and using this embedding by learning how to map images into it. Our new approach improves the performance of image-only low-shot learning approaches on multiple datasets. We also introduce Toys4K, a 3D object dataset with the largest number of object categories currently available, which supports low-shot learning.