Worker Activity Recognition
A leading construction company wanted to track efficiency of workers on a large construction site. The company wanted to measure if and for how long certain activity (e.g. welding) is performed. In addition, company wanted to identify when potentially dangerous actions are performed and ensure a safe working environment.
EasyFlow solution was integrated with security cameras at the construction site. Artificial intelligence and Convolutional Neural Networks were used to identify prerequisite worker activities. The solution was able to adapt to different lighting conditions. It also took into account the lower resolution of some surveillance cameras, as well overlapping objects which restrict view.
The client could identify the duration of predetermined worker activities. Such evaluations allowed to measure productivity across worker teams, compare their performance and identify best-performing workers. These insights were incorporated into better time management for projects. In addition, it helped to identify potentially dangerous activities and reduce the risk of workplace accidents.