Compact Bilinear Pooling

TL;DR

This paper introduces two compact bilinear pooling methods that retain the discriminative power of full bilinear features while significantly reducing dimensionality, enabling efficient end-to-end training for visual recognition tasks.

Contribution

The paper presents novel compact bilinear pooling techniques derived from a kernelized analysis, improving efficiency without sacrificing accuracy in visual recognition.

Findings

Achieves comparable performance to full bilinear features with only a few thousand dimensions.

Enables end-to-end training through back-propagation of classification errors.

Effective for image classification and few-shot learning across multiple datasets.

Abstract

Bilinear models has been shown to achieve impressive performance on a wide range of visual tasks, such as semantic segmentation, fine grained recognition and face recognition. However, bilinear features are high dimensional, typically on the order of hundreds of thousands to a few million, which makes them impractical for subsequent analysis. We propose two compact bilinear representations with the same discriminative power as the full bilinear representation but with only a few thousand dimensions. Our compact representations allow back-propagation of classification errors enabling an end-to-end optimization of the visual recognition system. The compact bilinear representations are derived through a novel kernelized analysis of bilinear pooling which provide insights into the discriminative power of bilinear pooling, and a platform for further research in compact pooling methods. Experimentation illustrate the utility of the proposed representations for image classification and few-shot learning across several datasets.