Related Experiment Video
Updated: Jun 25, 2025
07:30
A Test Bed to Examine Helmet Fit and Retention and Biomechanical Measures of Head and Neck Injury in Simulated Impact
Published on: September 21, 2017
8.9K
Lightweight helmet target detection algorithm combined with Effici-Bi-Level Routing Attention
Yanguo Huang1, Minjie Fang1, Jian Peng1
1College of Electrical and Automation, Jiangxi University of Science and Technology, GanZhou, China.
Plos One
|May 29, 2024
Summary
FB-YOLOv7 enhances helmet detection accuracy by improving small target identification and reducing errors. This advanced network offers high efficiency and superior performance compared to existing models for two-wheeler safety.
Area of Science:
- Computer Vision
- Artificial Intelligence
- Road Safety
Background:
- Helmet use is crucial for reducing two-wheeler accident injuries.
- Existing helmet detection systems struggle with small targets and equipment limitations, leading to missed or false detections.
Purpose of the Study:
- To develop an improved helmet detection network, FB-YOLOv7, addressing limitations of existing models.
- Enhance the accuracy and efficiency of detecting helmets in real-world traffic scenarios.
Main Methods:
- Utilized YOLOv7-tiny as a base and introduced an enhanced Bi-Level Routing Attention mechanism.
- Implemented the AFPN framework with asymptotic adaptive feature fusion and EfficiCIoU loss for improved accuracy.
Main Results:
- FB-YOLOv7 achieved 87.2% and 94.6% mean average precision (mAP@.5).
- Demonstrated high efficiency with frame rates of 129 and 126 frames per second (FPS).
- Outperformed six other detection networks in accuracy, small target sensitivity, and efficiency.
Conclusions:
- FB-YOLOv7 significantly improves helmet detection performance, particularly for small targets.
- The network offers a practical and efficient solution for enhancing road safety through better helmet compliance monitoring.
