Helmet wear detection based on neural network algorithm

Abstract

Wearing safety helmet correctly is an effective means to reduce the accident rate and ensure the safety of construction personnel. But in the complex work site, the helmet detection task will face a variety of challenges. This paper uses YOLOV4's deep neural network architecture and makes fine-tuning and improvement according to the characteristics and difficulties of this task, proposes a new helmet detection system, and achieves 95.1% accuracy. The application can timely and accurately detect whether the personnel wear the helmet correctly, which is of great significance for ensuring construction safety in practical work.