A Holistic Work Zone Safety Alert System through Automated Video and Smartphone Sensor Data Analysis (05-089)

Project Description: The project aimed to address the gap in existing work zone safety monitoring systems by leveraging advanced technologies to track workers, identify hazardous situations, and alert workers of danger. The focus was on both external safety hazards (e.g., accidents due to vehicular intrusions) and internal safety risks (e.g., improper use of vibrating hand tools). The project used a holistic approach that involved analyzing visual data from surveillance cameras and utilizing wearable sensor data for safety monitoring and alert generation. In terms of data collection, the project collected data from the job site of Casper company on 9th June 2022. It employed smartphones as a ubiquitous platform for data collection and communication. A mobile application was developed to generate alerts when unsafe vibration levels were detected in proximity to high vibration power tools, such as jackhammers. In addition to the data collected from the job site of Casper company, a deep learning model was utilized in the project to classify and detect speed of vehicles in video frames. These video frames were captured by traffic cameras installed at an intersection located at H St and Broadway in the City of Chula Vista during March 2022. The deep learning model was applied to analyze the visual data and identify any relevant objects or events that could pose safety risks within the work zone. This component of the project aimed to enhance the overall work zone safety monitoring system by leveraging advanced computer vision techniques for real-time object detection and classification. Data Scope: The data set collected for this project from right arm consists of 46,000 rows of data, capturing the acceleration of the worker's arms in the x, y, and z directions. The data was collected using smartphones attached to both of the worker's arms during work zone activities involving the TE 1000 AVR and HM1203C power tools. , ensuring redundancy in data collection. The accelerometer data was stored in the server and recorded at a sampling frequency of 500 Hz. Additional information can be found in the the Safe-D_05-089_Data_Description document below. Data Specification: The data specification information can be found in the the Safe-D_05-089_Data_Description document below.