Human Activity Recognition in Maintenance Centers to Reduce Wasted Time

This paper proposed a system that extracts workers' poses from live cam and video clips using mode classification. In this paper, we tested two algorithms to detect worker activity. This system aims to detect and classify the positive and negative activities of the worker in car maintenance centers such as (changing the tire, changing the oil, using the phone, standing without work) by calculating the time of each activity to measure the worker's performance and determining the lost time correctly and properly. Two experiments were conducted, the first experiment was conducted to measure the performance of the dollar algorithm with different participants. The results showed that the 1 dollar recognizer achieved 94.2% accuracy when tested on 364 different videos. The second experiment was conducted to measure the accuracy of the system in recognizing real-time activities from the live camera. It was conducted on 5 participants in a controlled environment. The system achieved an accuracy rate of 93.3%.