Tracking workers' productivity in real-time using computer vision
| Author(s) | Kadoura, Ahmad | |
| Date Accessioned | 2021-12-01T17:54:19Z | |
| Date Available | 2021-12-01T17:54:19Z | |
| Publication Date | 2021 | |
| SWORD Update | 2021-08-11T16:05:46Z | |
| Abstract | The international construction industry exceeds $10 Trillion annually making it one of the largest industries across the globe. Technology has transformed the tools for construction project delivery and material advances continue to improve life-cycle performance. Yet these advancements have not resulted in increased output and the construction industry has remained stagnant in term of productivity growth for decades. This productivity ‘crisis’ adversely impacts the industry by reducing the potential and also impacts individual projects through increased costs and completion times. Research was initiated to make a difference in this area with the goal of developing better management tools to improve the productivity of workers and crews in a real-time fashion. Productivity, where studied in construction, is typically done forensically after the activities are complement. The goal of this effort is to develop an approach whereby real-time or near real-time feedback could be provided when corrective action can impact the outcome. The approach employed focuses on developing an algorithm capable of providing real-time feedback on the productivity of workers using Computer Vision models and construction site videography. The models developed are based on simple classifier that checks whether a worker is performing work or standing idle. Where significant non-work periods are identified, corrective action can then be taken. The algorithm developed only addresses labor productivity. Equipment productivity is included using a simplified metric based on whether the equipment is under load, idling or turned off. The research demonstrates effectiveness of the techniques through hypothetical situations. Further research, which can expand, refine and further test the algorithms, are outlined in conclusion. | en_US |
| Advisor | Small, Edgar P. | |
| Degree | M.C.E. | |
| Department | University of Delaware, Department of Civil and Environmental Engineering | |
| DOI | https://doi.org/10.58088/g3da-eg61 | |
| Unique Identifier | 1286887892 | |
| URL | https://udspace.udel.edu/handle/19716/29482 | |
| Language | en | |
| Publisher | University of Delaware | en_US |
| URI | https://login.udel.idm.oclc.org/login?url=https://www.proquest.com/dissertations-theses/tracking-workers-productivity-real-time-using/docview/2572581107/se-2?accountid=10457 | |
| Keywords | Computer Vision | en_US |
| Keywords | Construction productivity | en_US |
| Keywords | Worker productivity | en_US |
| Keywords | Real-time tracking | en_US |
| Title | Tracking workers' productivity in real-time using computer vision | en_US |
| Type | Thesis | en_US |