A Real-Time High-Accuracy Framework for Small-Scale PPE Detection in Complex Construction Environments

Accurate detection of personal protective equipment (PPE) is recognized as crucial for ensuring construction site safety and regulatory compliance. However, small-scale PPE objects under challenging conditions such as long distances, occlusions, and cluttered backgrounds are frequently mismanaged by existing systems. This study proposes construction site personal protective equipment YOLO (CSPPE-YOLO), an enhanced object detection framework based on YOLOv11, to address the significant challenges for small-scale PPE detection in complex construction environments. CSPPE-YOLO incorporates three core architectural enhancements: the Focaler-EIoU loss function to mitigate class imbalance and improve localization precision for challenging objects, the feature fusion perceptual module (FFPM) to strengthen multiscale feature representation for small-target detection, and the router feature pyramid network (Router-FPN) architecture to enable adaptive feature pyramid optimization across diverse object scales. To further strengthen small-target detection capabilities, this study introduces CSPPE-YOLO-p2, an advanced variant incorporating an additional p2 detection layer specifically designed for ultrasmall PPE objects with enhanced precision. Experimental evaluation on the site object detection data set (SODA) demonstrates that CSPPE-YOLO significantly outperforms current state-of-the-art methods, achieving a mAP50 of 71.5% while maintaining real-time inference at 52.4 frames per second, representing substantial improvements in both detection accuracy and computational efficiency. The enhanced CSPPE-YOLO-p2 variant further elevates performance, achieving a mAP50 of 79.9% and precision of 83.7%, validating its superior capability for detecting small-scale PPE objects in complex construction environments. These results demonstrate the effectiveness of the proposed framework in addressing critical challenges of PPE detection and compliance monitoring in complex construction sites, particularly for small-scale objects that are difficult to identify under challenging environments. This delivers a reliable and efficient automated solution for PPE compliance verification and real-time safety management, empowering construction industry stakeholders to implement proactive safety protocols while enhancing overall site safety standards and ensuring regulatory compliance.