Construction labor activity recognition via an imu-implemented wristband

Human labor is a pivotal element in determining project costs and overall success within the construction industry, necessitating robust monitoring of workers' productivity, well-being, and the cultivation of a sufficient and constructive association between workers and employers. The initial step in productivity assessment is the accurate identification of workers' activities. Accordingly, this research focuses on recognizing and classifying construction workers' activities as a critical step toward improving productivity and fostering fairness within the industry. This study presents a novel framework for classifying construction workers' activities using a custom-designed wristband embedded with an inertial measurement unit (IMU), comprising an accelerometer and a gyroscope. A one-dimensional convolutional neural network (1D-CNN) model was developed to process the captured data and accurately classify various activities. The activities examined included masonry, tiling, painting, shoveling, and sawing. The proposed approach demonstrated an accuracy exceeding 99% under controlled laboratory conditions and achieved 96% accuracy in identifying painting activity in simulated real-world conditions. These findings, marked by both high precision and the diverse range of activities recognized, highlight the robustness of the proposed method and provide a strong foundation for future research in this field.