Integrating real-time pose estimation and PPE detection with cutting-edge deep learning for enhanced safety and rescue operations in the mining industry

• Published in: Neurocomputing (Amsterdam) Vol. 618; p. 129080
• Main Authors: Imam, Mohamed, Baïna, Karim, Tabii, Youness, Ressami, El Mostafa, Adlaoui, Youssef, Boufousse, Soukaina, Benzakour, Intissar, Abdelwahed, El Hassan
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 14.02.2025
• Subjects: Deep learning; Human pose estimation (HPE); Machine learning; Miners’ safety; Personal Protective Equipment (PPE); Underground mines; Underground mines; Deep learning; Human pose estimation (HPE); Personal Protective Equipment (PPE); Miners’ safety; Machine learning
• ISSN: 0925-2312
• DOI: 10.1016/j.neucom.2024.129080

The dynamic and often hazardous environments of mines necessitate advanced safety protocols, particularly in PPE compliance and emergency response. This study leverages cutting-edge pose estimation and object detection technologies to enhance miner safety by focusing on two primary goals: developing a comprehensive dataset for body situation classification from miner pose data to improve rescue operations, and refining PPE detection using deep learning models localized by pose keypoints. A robust dataset was collected in diverse mining scenarios under low-light and noisy conditions using the adapted YOLO Pose v8. This dataset was instrumental in extracting detailed metrics such as joint angles and bounding box sizes for body situation classification. For PPE detection, the study employed various YOLO models, including YOLOv8, YOLOv9, YOLOv10, YOLOv11, YOLO World, and RTDetr, across a public dataset and a custom dataset from five Moroccan mines to ensure effective PPE identification in different contexts. The implemented models showed superior performance in PPE detection and pose estimation, confirming the efficacy of integrating advanced YOLO architectures in mining operations. By enhancing detection and analyzing miner postures and PPE compliance, this research significantly improves mine safety, potentially reducing accident rates and improving rescue operation effectiveness. The methodologies and insights extend to occupational safety in other high-risk industries, setting new standards for safety technology integration and positioning these solutions at the forefront of innovation.