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.
@article{arxiv.2202.13080,
title = {Improved Hard Example Mining Approach for Single Shot Object Detectors},
author = {Aybora Koksal and Onder Tuzcuoglu and Kutalmis Gokalp Ince and Yoldas Ataseven and A. Aydin Alatan},
journal= {arXiv preprint arXiv:2202.13080},
year = {2022}
}
Comments
ICIP 2022. 5 pages, 2 figures, 7 tables. The codes are available at https://github.com/aybora/yolov5Loss