A Review of the Application of Machine Learning and Computer Vision Algorithms in Identifying, Assessing, and Mitigating Hazards in Construction Site Setup: An Analysis of Research Trends and Future Directions
A Review of the Application of Machine Learning and Computer Vision Algorithms in Identifying, Assessing, and Mitigating Hazards in Construction Site Setup: An Analysis of Research Trends and Future Directions
Dariush Jamali1 Alireza Valipour2
1) M.Sc. in Civil Engineering – Construction Engineering and Management, Shiraz Branch, Islamic Azad University, Shiraz, Iran.
2) Associate Professor, Civil Engineering – Construction Management, Shiraz Branch, Islamic Azad University, Shiraz, Iran
IOI Code
:
ABFG-ACAGGG
Publication :
12th.international Congress on Civil Engineering, Architecture, Art and Urban Development(12icsau.ir)
Abstract :
Safety management during the site mobilization phase of construction projects has consistently presented significant challenges, stemming from the dynamic nature of activities, the simultaneous presence of machinery and personnel, and the constant reconfiguration of the site layout. Traditional, static approaches—relying on qualitative judgments and periodic inspections—have proven inadequate for addressing the complexities inherent in modern projects. This research aims to provide a comprehensive examination of the application of machine learning and computer vision algorithms in identifying, assessing, and mitigating risks during site mobilization, while also analyzing research trends and future directions. To this end, this study—based on a systematic review of scientific literature and existing research data—analyzes the role of smart technologies within the risk management cycle. The findings indicate that computer vision systems are highly effective for real-time environmental monitoring and the automated detection of incidents, such as machinery interference or the falling of construction materials. Furthermore, integrating artificial intelligence models with risk assessment tools (e.g., FMEA) and employing evolutionary optimization algorithms (such as genetic algorithms) to optimize site layout enables the dynamic calculation of the Risk Priority Number (RPN) and a substantial reduction in hazards; indeed, analyses demonstrate a significant decrease in RPN values for critical risks. Nevertheless, the practical implementation of these technologies faces several obstacles, including data quality issues, deployment costs, and organizational resistance. The conclusion indicates that artificial intelligence is not a substitute for human judgment; rather, as a proactive tool and decision-support aid, it plays a key role in the transition from reactive management to systematic, data-driven safety.
Keywords : safety management site setup machine vision machine learning risk assessment