Maximum-Entropy Fine-Grained Classification

TL;DR

This paper introduces a maximum-entropy training routine for fine-grained visual classification that improves robustness and achieves state-of-the-art results across various FGVC tasks.

Contribution

It revisits maximum-entropy learning for FGVC, providing a new training routine with theoretical and empirical support that enhances performance and robustness.

Findings

Achieves state-of-the-art performance on multiple FGVC benchmarks.

Robust to hyperparameter variations, training data quantity, and label noise.

Provides a theoretically justified training routine for fine-grained classification.

Abstract

Fine-Grained Visual Classification (FGVC) is an important computer vision problem that involves small diversity within the different classes, and often requires expert annotators to collect data. Utilizing this notion of small visual diversity, we revisit Maximum-Entropy learning in the context of fine-grained classification, and provide a training routine that maximizes the entropy of the output probability distribution for training convolutional neural networks on FGVC tasks. We provide a theoretical as well as empirical justification of our approach, and achieve state-of-the-art performance across a variety of classification tasks in FGVC, that can potentially be extended to any fine-tuning task. Our method is robust to different hyperparameter values, amount of training data and amount of training label noise and can hence be a valuable tool in many similar problems.