Computational method for monitoring pauses exercises in office workers through a vision model

Abstract

A sedentary routine at work can cause various muscular, skeletal or visual diseases, however these can be prevented with what is known as active pauses. This article is intended to illustrate how software can help reduce the risk of occupational disease due to the sedentary lifestyle of an office job, for this purpose a web application was developed under the SCRUM methodology, which makes use of the TensorFlow, Flask PoseNet model and python, for an active pause control application which is a proven practice of reducing the type of diseases already mentioned. With these tools it was possible to develop an algorithm capable of comparing two human figures; which serves to compare whether the user of the program is performing or not correctly performing the active pause exercise, with an average error squared on the order of 10−32. Finally, The application can keep track of the figure and exercises performed by the user just by using the user personal webcam and the comparison algorithm developed, leaving behind the use of tools such as Kinect.