Lightweight YOLO v4 security protection detection method based on attention mechanism improvement

Abstract

The invention discloses a lightweight YOLO v4 security protection detection method based on attention mechanism improvement, and the method comprises the steps: collecting images, building a data set, employing a YOLO v4 network as a basic model, integrating an attention mechanism into a YOLO v4 model, and obtaining an improved YOLO v4 security protection detection model; training the improved YOLO v4 safety protection detection model by using the data set to obtain an improved YOLO v4 target detection model; inputting a to-be-detected image into the improved YOLO v4 safety protection detection model, outputting corresponding target detection results by adopting the improved YOLO v4 target detection model, respectively positioning a pedestrian detection frame, a safety helmet detection frame and a safety reflective vest detection frame, and respectively calculating CIoU of three targets to obtain a final target detection frame; and detecting and identifying a construction scene safety equipment target, and judging whether a behavior of irregularly wearing the safety equipment exists in the image or not according to the coincidence condition of the target pedestrian, the safety helmet and the safety reflective vest detection frame.

Description

Technical field [0001] The present invention belongs to the field of computer visual technology, and specifically, a lightweight YOLO V4 safety protection detection method based on attention mechanism, which is applied to various types of construction site for real-time safety cap wearing testing and safety protection testing. Background technique [0002] In recent years, China's video surveillance technology has achieved great development. The 24 / 7 surveillance cameras have basically allocated in all corners of the city, which greatly guarantees the safety of the people's lives and property. Considering the shortcomings of manual monitoring, the existing scholars propose combined with computer vision technologies and apply them to the construction site, gradually replace artificial screening and monitoring, enabling it to improve the efficiency of security monitoring. [0003] In combination with the objective detection of deep learning, video surveillance technology based on a...

AI Technical Summary

Problems solved by technology

[0003] Combined with the method of target detection in deep learning, the video surveillance technology based on all-weather monitoring has been widely used in the fields of intelligent video surveillance and tracking, flow control of people and vehicles, security maintenance of public areas, etc., but the application in the detection of safety protection equipment but less

Images