USING COMPUTER VISION TO IMPROVE CONSTRUCTION SAFETY: A FEASIBILITY STUDY

Abstract

Past studies have highlighted the applications of computer vision and discussed its benefits and challenges. Despite such efforts, computer vision is still not widely utilised in the construction industry. Today, the construction industry contributes to majority of the fatalities and accidents in Singapore and the world. Singapore has a target to reduce the workplace fatality rate to under 1.0 per 100,000 workers by 2028. To achieve this goal, there is a need to solve this safety gap in the construction industry, and this can be done with vision-based monitoring. This study has found that there are several benefits and challenges to implementing computer vision in the construction industry. Based on its applications, it can potentially improve the industry’s safety level. Following which, the feasibility study has utilised a cost-benefit analysis and established that computer vision is indeed feasible. A table highlighting the different commercial computer vision developers which offer their services has also been crafted. In a bid to encourage main contractors to implement computer vision, this table has presented the availability of this technology to the public. It has found that despite its high cost, the benefits are magnanimous and vision-based monitoring can potentially automate the manual process of monitoring the construction site and fill the industry’s safety gap. To better justify the feasibility of vision-base monitoring, an action research on a pilot study has also been performed. The results of this study have proven that computer vision is able to detect the construction workers’ helmets and identify workers who have entered the more dangerous zones at the construction site. This dissertation has also suggested several recommendations to improve on the technology’s challenges and limitations. Besides these specific solutions, overall recommendations have been made. One key recommendation is to integrate vision-based monitoring with other forms of technology such as Unmanned Aerial Vehicles (UAVs) or wearables, allowing the system to be more wholesome and holistic. The paper has also discussed other considerations for firms when implementing vision-based monitoring as the management should take note of how the changes affect the workers and continue improving its overall workplace safety management system.