RL-YOLO:改进YOLOv8n的安全帽佩戴检测算法

Journal: Advances in Computer and Autonomous Intelligence Research DOI: 10.12238/acair.v2i4.10357

杨垚磊1, 阮景奎1, 王宸1, 闫伟伟2

1. 湖北汽车工业学院 机械工程学院
2. 驰田汽车股份有限公司

Abstract

在工业生产中,佩戴安全帽是对施工人员头部安全的一项重要措施。针对改装车间工人安全帽佩戴不规范、设备繁多、人员工作密集等特点,设计基于YOLOv8n模型的安全帽佩戴检测算法。算法推出重校准金字塔网络(Re-CalibrationFPN)改进原模型的颈部网络,新增小目标特征信息层,加强了模型的深层与浅层交互能力,从而增加模型检测的准确率;同时使用轻量化的检测头LSD(Lightweight Shared Detection Head),通过使用共享卷积的方式减少模型参数量与计算量,使模型便于安装部署到终端;结合使用Wise-IoU和Powerful-IoUv2的思想改进损失函数,引入新的惩罚机制措施,提升模型处理复杂样本的能力。实验结果表明,改进后的RCLS-YOLOv8模型在公开数据集SHWD(Safety Helmet Wearing Dataset)上P降低了0.2%,R、mAP@0.5分别提升了3.9%和2.9%,能够满足检测和部署要求。

Keywords

YOLOv8;实时监测;重校准金字塔网络;LSD检测头;损失函数

References

[1] HOSHINO W,SEO J,YAMAZAKI Y.A study for detecting disa ster victims using multi- copter drone with a thermographic camera and image object recognition by SSD[C]// 2021 IEEE/ASME International Conference on Advanced Intelligent Mechatronics(AIM),2021:162-167.
[2] SORBELLI F B, PALAZZETTI L, PINOTTI C M. YOLObased det ection of halyomorpha halys in orchards using RGB cameras anddrones[J]. Computers and Electronics in Agriculture, 2023, 213:108228.
[3] PIRASTEH S, RASHIDI P, RASTIVEIS H, et al. Developing an algorithm for buildings extraction and determining changes from airborne LiDAR, and comparing with R-CNN method from drone images[J]. Remote Sensing,2019,11(11): 1272.
[4] SEO D M, WOO H J, KIM M S, et al.Identification of asbestos slates in buildings based on faster region-based convolutional neural network (Faster R-CNN) and drone-based aerial imagery[J]. Drones, 2022,6(8):194.
[5] YANG S D,CHEN Z H, MA X M,etal.Real-time high-precisi on pedestrian tracking: a detection-tracking-correction stra tegy based on improved SSD and cascade R-CNN[J]. Journal of Real-Time Image Processing,2022,19(2):287-302.
[6] GUO C S, CAI M,YING N,etal.ANMS:attention-based nonmaximum suppression[J].Multimedia Tools and Applications, 2022,81(8):11205-11219.
[7] YANG W, ZHOU G L, GU Z W,etal.Safety helmet wearing det ection based on an improved YOLOv3 scheme[J]. International Journal of Innovative Computing, Information and Control, 2022,18(3):973-988.
[8] WANG J R,LI S M,AN Z H,etal.Batch-normalized deep neur al networks for achieving fast intelligent fault diagnosis of machines[J]. Neurocomputing,2019,329:53-65.
[9] 宋晓凤,吴云军,刘冰冰,等.改进YOLOv5s算法的安全帽佩戴检测[J].计算机工程与应用,2023,59(2):194-201.
[10] 杨大为,张成超.基于改进YOLOv7的安全帽佩戴检测算法[J].沈阳理工大学学报,2024,43(01):16-21.7.
[11] Dong S, He Q,Li H,etal.Automated PPE misuse identifi cation and assessment for safety performance enhancement[M]// ICCREM2015.2015:204-214.
[12] 戚玲珑,高建瓴.基于改进YOLOv7的小目标检测[J].计算机工程,2023,49(1):41-48.
[13] LI Y,FAN Q,HUANG H, et al. A modified YOLOv8 detection network for UAV aerial image recognition[J].Drones,2023,7(5): 304.
[14] 潘玮,韦超,钱春雨,杨哲.面向无人机视角下小目标检测的YOLOv8s改进模型[J].计算机工程与应用,2024,60(9):142-150.
[15] Feilong Tang,Qiming Huang,Jinfeng Wang,Xianxu Hou, Jionglong Su,Jingxin Liu.DuAT: Dual-Aggregation Transformer Network for Medical Image Segmentation
[16] Tong, Z.,Chen,Y.,Xu,Z.,&Yu,R. (2023). Wise-iou: Bounding box regression loss with dynamic focusing mechanism. arXiv preprint arXiv:2301.10051.
[17] Liu C,Wang K,Li Q,etal.Powerful-IoU:More straightfor ward and faster bounding box regression loss with a nonmon otonic focusing mechanism[J].Neural Networks,2024,170:276- 284.
[18] Deng-Ping Fan,Ge-Peng Ji,Tao Zhou,Geng Chen,Huazhu Fu,Jianbing Shen,and Ling Shao. Pranet:Parallel reverse atten tion network for polyp segmentation. In Internationalconfere nce on medical image computing and computer assisted interv ention,pages263-273.Springer,2020.

Copyright © 2024 杨垚磊, 阮景奎, 王宸, 闫伟伟

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License