Joint Learning of Generative Translator and Classifier for Visually Similar Classes

TL;DR

This paper introduces GTCN, a joint learning framework that combines generative translation and classification to enhance visual recognition of similar classes with limited data, outperforming baselines and matching state-of-the-art results.

Contribution

The paper presents a novel end-to-end joint learning approach for generative translation and classification, incorporating adaptive loss functions for improved data augmentation in visual recognition.

Findings

Training on 40% of data surpasses full-data baselines.

Achieves state-of-the-art performance with a lightweight architecture.

Effective data augmentation improves classification accuracy.

Abstract

In this paper, we propose a Generative Translation Classification Network (GTCN) for improving visual classification accuracy in settings where classes are visually similar and data is scarce. For this purpose, we propose joint learning from a scratch to train a classifier and a generative stochastic translation network end-to-end. The translation network is used to perform on-line data augmentation across classes, whereas previous works have mostly involved domain adaptation. To help the model further benefit from this data-augmentation, we introduce an adaptive fade-in loss and a quadruplet loss. We perform experiments on multiple datasets to demonstrate the proposed method's performance in varied settings. Of particular interest, training on 40% of the dataset is enough for our model to surpass the performance of baselines trained on the full dataset. When our architecture is trained on the full dataset, we achieve comparable performance with state-of-the-art methods despite using a light-weight architecture.