AI Framework Spots Unsafe Miner Behavior in Real Time with Dual Attention Pose Network

Researchers have developed MIUBM-Net, a dual-module AI framework that detects unsafe miner behaviors and posture with high accuracy and real-time speed under complex underground mining conditions. The system combines STFNet-A for standard unsafe actions and Alpha-SDL-Pose with Dual Non-Local Attention to monitor risky postures, addressing a major gap with a dedicated miner behavior dataset.

This Wire brief sits within Fusion42's coverage of Computer Vision, Industrial Inspection and AI Infrastructure.