A method for electric vehicle headgear wearing detection based on improved YOLOv8

CN122244791APending Publication Date: 2026-06-19CHANGZHOU UNIV

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Applications(China)
Current Assignee / Owner: CHANGZHOU UNIV
Filing Date: 2026-03-18
Publication Date: 2026-06-19

AI Technical Summary

Technical Problem

The existing YOLOv8 algorithm has problems in detecting helmet wearing on electric vehicles, such as insufficient ability to detect extremely small targets, insufficient feature extraction ability, and lack of attention mechanism. This results in high false negative rate and high false positive rate, making it difficult to achieve high-precision detection in complex traffic scenarios.

Method used

The C2FCIB module replaces the C2F module of the YOLOv8 backbone network, embeds the DBCA deformable two-level channel attention module, and adds a 160×160 scale micro-target detection head in the feature pyramid network to enhance feature extraction capabilities and adaptive attention mechanism.

Benefits of technology

It significantly improves the accuracy and robustness of electric vehicle helmet wearing detection, reduces the false negative rate, increases the recall rate for small targets and the average accuracy under multi-scale cross-connection ratio, and is suitable for intelligent urban traffic monitoring.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure CN122244791A_ABST

Patent Text Reader

Abstract

This invention relates to the field of image target detection technology, and provides an improved YOLOv8-based method for detecting helmet wearing on electric vehicles. The method includes: acquiring images of electric vehicle drivers; constructing and enhancing a helmet wearing detection dataset; constructing an improved YOLOv8 electric vehicle helmet wearing detection model: replacing the C2F modules in each stage of the YOLOv8 backbone network with C2FCIB modules; embedding a DBCA deformable two-level channel attention module after the output of each stage of the backbone network; adding a 160×160-scale micro-target detection head to the feature pyramid network; inputting the enhanced dataset into the improved YOLOv8 electric vehicle helmet wearing detection model for training; and using the trained model to detect the helmet wearing status of electric vehicle drivers, outputting the wearing status and location information. This invention improves the detection accuracy and recall rate of small-scale helmet targets in complex traffic scenarios, and reduces false negatives and missed detections.

Need to check novelty before this filing date? Find Prior Art