Improving Crowded Object Detection via Copy-Paste

TL;DR

This paper introduces a copy-paste data augmentation method to improve crowded object detection by addressing IoU-confidence correlation disturbances and confused de-duplication, achieving over 2% improvement in detection accuracy.

Contribution

The paper proposes a novel copy-paste augmentation scheme and a consensus learning method to enhance crowded object detection without additional labeling costs.

Findings

Improves state-of-the-art detectors by over 2% in crowded scenarios.

Outperforms existing data augmentation strategies for crowded detection.

Addresses key issues of IoU-confidence correlation and de-duplication in crowded scenes.

Abstract

Crowdedness caused by overlapping among similar objects is a ubiquitous challenge in the field of 2D visual object detection. In this paper, we first underline two main effects of the crowdedness issue: 1) IoU-confidence correlation disturbances (ICD) and 2) confused de-duplication (CDD). Then we explore a pathway of cracking these nuts from the perspective of data augmentation. Primarily, a particular copy-paste scheme is proposed towards making crowded scenes. Based on this operation, we first design a "consensus learning" method to further resist the ICD problem and then find out the pasting process naturally reveals a pseudo "depth" of object in the scene, which can be potentially used for alleviating CDD dilemma. Both methods are derived from magical using of the copy-pasting without extra cost for hand-labeling. Experiments show that our approach can easily improve the state-of-the-art detector in typical crowded detection task by more than 2% without any bells and whistles. Moreover, this work can outperform existing data augmentation strategies in crowded scenario.