Optimizing Helmet Use Detection in Construction Sites via Fuzzy Logic-Based State Tracking
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.This study introduces an advanced AI system for construction site safety, accurately detecting helmets and tracking workers even in difficult conditions. The innovative approach significantly reduces errors and improves tracking stability for real-time monitoring.
Area Of Science
- Computer Vision
- Artificial Intelligence
- Construction Safety
Background
- Automated safety monitoring on construction sites is crucial.
- Challenges include precise helmet detection and multi-object tracking in complex video sequences with occlusions.
Purpose Of The Study
- To develop a robust two-stage framework for helmet-status detection and multi-object tracking.
- To enhance accuracy and reduce identity fragmentation in challenging construction site videos.
Main Methods
- A YOLOv5 detector enhanced with self-adaptive coordinate attention (SACA) for improved head-region cue emphasis and small-object discrimination.
- A DeepSORT tracker with fuzzy-logic gating and temporally consistent update rules to stabilize trajectories and minimize identity fragmentation.
Main Results
- The SACA-enhanced YOLOv5 achieved a mAP@0.5 of 0.940, outperforming YOLOv8 and YOLOv9.
- The tracker achieved 90.5% MOTA and 84.2% IDF1 with only five identity switches.
- The system demonstrated effectiveness in reducing missed detections and identity fragmentation under occlusion, varied illumination, and camera motion.
Conclusions
- The proposed framework provides accurate and stable helmet monitoring for construction sites.
- The system is suitable for real-time deployment in complex and challenging environments.
- SACA modules and temporal consistency rules are key to the system's performance improvements.
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
