Improving filling level classification with adversarial training

TL;DR

This paper enhances filling level classification in images of cups by employing adversarial transfer learning, which improves accuracy and reduces overfitting despite limited training data.

Contribution

It introduces a transfer learning approach using adversarial training on a source dataset, refined with task-specific data, to improve classification of container fill levels.

Findings

Transfer learning with adversarial training improves accuracy.

The method reduces overfitting to training data.

Experimental results on the CORSMAL dataset validate effectiveness.

Abstract

We investigate the problem of classifying - from a single image - the level of content in a cup or a drinking glass. This problem is made challenging by several ambiguities caused by transparencies, shape variations and partial occlusions, and by the availability of only small training datasets. In this paper, we tackle this problem with an appropriate strategy for transfer learning. Specifically, we use adversarial training in a generic source dataset and then refine the training with a task-specific dataset. We also discuss and experimentally evaluate several training strategies and their combination on a range of container types of the CORSMAL Containers Manipulation dataset. We show that transfer learning with adversarial training in the source domain consistently improves the classification accuracy on the test set and limits the overfitting of the classifier to specific features of the training data.