Improved Hard Example Mining Approach for Single Shot Object Detectors

TL;DR

This paper enhances single shot object detectors by combining and adapting hard example mining techniques, leading to significant improvements in detection accuracy on challenging datasets.

Contribution

It introduces a novel combined hard example mining approach integrated into YOLOv5, improving detection performance over existing methods.

Findings

Increases mAP by 3% over original loss

Achieves 1-2% higher mAP than individual hard-mining methods

Demonstrates effectiveness on Anti-UAV Challenge Dataset

Abstract

Hard example mining methods generally improve the performance of the object detectors, which suffer from imbalanced training sets. In this work, two existing hard example mining approaches (LRM and focal loss, FL) are adapted and combined in a state-of-the-art real-time object detector, YOLOv5. The effectiveness of the proposed approach for improving the performance on hard examples is extensively evaluated. The proposed method increases mAP by 3% compared to using the original loss function and around 1-2% compared to using the hard-mining methods (LRM or FL) individually on 2021 Anti-UAV Challenge Dataset.