Online Anchor-based Training for Image Classification Tasks

TL;DR

This paper introduces Online Anchor-based Training (OAT), a novel method for image classification that trains models to learn label changes relative to anchors, improving performance across multiple datasets.

Contribution

The paper proposes a new anchor-based training approach inspired by object detection, shifting from direct label learning to relative label changes for improved accuracy.

Findings

Validated on four datasets showing performance gains

Outperforms traditional training methods

Effective in diverse image classification tasks

Abstract

In this paper, we aim to improve the performance of a deep learning model towards image classification tasks, proposing a novel anchor-based training methodology, named \textit{Online Anchor-based Training} (OAT). The OAT method, guided by the insights provided in the anchor-based object detection methodologies, instead of learning directly the class labels, proposes to train a model to learn percentage changes of the class labels with respect to defined anchors. We define as anchors the batch centers at the output of the model. Then, during the test phase, the predictions are converted back to the original class label space, and the performance is evaluated. The effectiveness of the OAT method is validated on four datasets.