Artificial Intelligence-based Safety Helmet Recognition on Embedded Devices to Enhance Safety Monitoring Process

Construction workers can be adequately protected by wearing a safety helmet while working. Due to the discomfort, the workers take off safety helmets while working, which is unsafe behavior and causes an injury or fatality in case of a fall. Therefore, a practical and handy solution is needed on the construction site to recognize workers safety helmets in order to determine their unsafe behavior. However, conventional safety monitoring methods are labor-intensive, time-consuming, and require a safety manager's presence, which is impossible for him to monitor all the construction workers performing different activities. Therefore, this research presented efficient and cost-effective Artificial Intelligence (Computer Vision) based mobile solution to monitor worker safety helmets and generate an alarming message to the safety manager and the workers. The proposed solution consists of (1) CV based object detection approach to recognize workers with and without a safety helmet, (2) deployment on edge devices such as Android smartphones (3) uses SMS'Manager API and ToneGenerator class to notify safety manager and worker, (4) and real-time firebase database to keep a record of the workers activities (safe and unsafe). Once the worker is detected without a safety helmet, the application generates and sends an SMS on the safety manager's cellphone with workers details and an audible alarm on the device speaker to make the worker aware of his unsafe action. The developed application will be extended with other case scenarios and include rewarding and penalising functionality based on records in the database.