南方科技大学知识苑(SUSTech KC): 基于YOLO改进模型的安全头盔佩戴检测的方法及系统

Title	基于YOLO改进模型的安全头盔佩戴检测的方法及系统
Alternative Title	Safety helmet wearing detection method and system based on YOLO improved model
Author	孟庆虎; 倪男; 胡超
First Inventor	孟庆虎
Original applicant	南方科技大学嘉兴研究院
First applicant	南方科技大学嘉兴研究院
Address of First applicant	314000 浙江省嘉兴市秀洲区王店镇吉蚂西路1号物流科技大楼3楼
Current applicant	南方科技大学嘉兴研究院
Address of Current applicant	314000 浙江省嘉兴市秀洲区王店镇吉蚂西路1号物流科技大楼3楼 (浙江,嘉兴,秀洲区)
First Current Applicant	南方科技大学嘉兴研究院
Address of First Current Applicant	314000 浙江省嘉兴市秀洲区王店镇吉蚂西路1号物流科技大楼3楼 (浙江,嘉兴,秀洲区)
Application Number	CN202111118606.1
Application Date	2021-09-24
Open (Notice) Number	CN113781469A
Date Available	2021-12-10
Status of Patent	实质审查
Legal Date	2021-12-28
Subtype	发明申请
SUSTech Authorship	First
Abstract	本发明提供一种基于YOLO改进模型的安全头盔佩戴检测的方法及系统。通过施工现场的监控图像获得训练图像和待检测图像，其中所述监控图像包括工作人员佩戴头盔的图像；将所述训练图像输入到基于YOLOv5结构改进的网络模型进行优化训练，得到改进模型；将所述待检测图像输入到所述改进模型并输出所述待检测图像中头盔佩戴的预测结果。本发明通过采取改进的CSPDarknet53提取目标特征，增加检测颈在YOLO基础上的特征融合的尺度，并在输出端增加了新的检测头以负责小目标的检测。通过获得先验尺寸特征，同时对损失函数和预测框筛选等指标进行了优化。提高改进模型对施工现场安全头盔佩戴检测的准确性，从而保证建筑施工的安全性。
Other Abstract	The invention provides a safety helmet wearing detection method and system based on a YOLO improved model. A training image and a to-be-detected image are obtained through a monitoring image of a construction site, wherein the monitoring image comprises an image that a worker wears a helmet; the training image is inputted into a network model improved based on a YOLOv5 structure for optimization training to obtain an improved model; and the to-be-detected image is inputted into the improved model and outputting a prediction result of helmet wearing in the to-be-detected image. According to the method, target features are extracted by adopting improved CSPDarknet53, the feature fusion scale of a detection neck on the basis of YOLO is increased, and a new detection head is added at an output end to be responsible for small target detection. The prior size features are obtained, and indexes such as a loss function and prediction frame screening are optimized at the same time. The accuracy of the improved model for detecting the wearing of the safety helmet on the construction site is improved so that the safety of building construction is ensured.
CPC Classification Number	G06T7/0002 ; G06N3/084 ; G06T2207/20081 ; G06N3/045 ; G06F18/23 ; G06F18/253
IPC Classification Number	G06T7/00 ; G06K9/62 ; G06N3/04 ; G06N3/08
INPADOC Legal Status	（ENTRY INTO FORCE OF REQUEST FOR SUBSTANTIVE EXAMINATION）[2021-12-28][CN]
INPADOC Patent Family Count	1
Extended Patent Family Count	1
Priority date	2021-09-24
Patent Agent	戴锦跃
Agency	深圳树贤专利代理事务所(普通合伙)
URL	[Source Record]
Data Source	PatSnap
Document Type	Patent
Identifier	http://kc.sustech.edu.cn/handle/2SGJ60CL/532838
Department	Department of Electrical and Electronic Engineering
Recommended Citation GB/T 7714	孟庆虎,倪男,胡超. 基于YOLO改进模型的安全头盔佩戴检测的方法及系统.

Files in This Item:
There are no files associated with this item.

If you have any objections to this item, please fill out the form below and the administrator will contact you as soon as possible.
Content:
Email：	*
Affiliation Sequential No:
Verification Code:	Refresh

Any comments and suggestions are welcomed.
Title:	*
Content:
Email：	*
Verification Code:	Refresh