Performance Evaluation of Deep Learning Based Industrial Safety Enhancement in Multiple Safety Equipment Detection

The different kinds of accidents that happen in construction site and industrial areas. Because many works carry a great compact of risk, it is imperative that industrial site personnel obey to complete safety protocols. The safety has always been a very main issue in all industrial activities, especially hazardous work. Workers are well equipped with safety equipment have less chance from accidents and injuries at industrial site. Safety equipment’s act as the first line of protection against injury, but they only work when they are wear in proper way. This work proposes a safety system to workers which include Hard Hats, Masks, Vest, Boost, Gloves etc., A system with real time object detection which is based on video streaming analysis and deep neural network techniques CNN, R-CNN to find the good object detection and classification. The several objects in a video source is given as the input to the system. An alarm is sent off when an individual step into the danger zone and their threshold value drops below. A method in convolution neural network forecast the workers wearing safety equipment in proper way or not. YOLO (You Only Look One) architecture is used in a deep learning-based object detection system to identify safety equipment at a single shot. By using this model, industrial areas can better operate for the safety of human life and the observing of worker safety.