Application of Python and OpenCV on industrial cycle time study

Abstract

Motion and time study enhance business performance by improving productivity. The cycle times are collected repetitively to confirm their accuracy and precision. The time taken to complete tasks by engineers or technicians varies due to work experience and educational background, which can cause a large number of repetitions in the time observation. Recent technologies to help shorten time and motion studies as process improvement include digital stopwatches and mobile applications. However, these only reduce documentation time, not observation time. Therefore, this project integrated machine vision technology to reduce observation time in a motion and time study project. The proposed algorithm was developed using OpenCV with Python. Cycle time, measured by the proposed algorithm using work process videos, was then compared with cycle time observed by human appraisers. Results confirmed that the cycle time detected by the proposed algorithm differed from the cycle time evaluated by appraisers with lower variation, i.e., requiring less replicates of observations. Outcomes of this research can be used to shorten the time study process and facilitate remote monitoring in process improvement projects.