A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Real-Time Ergonomic Risk Assessment Approach for Construction Workers Based on Computer Vision
https://orcid.org/http://orcid.org/0000-0003-4367-4219
https://orcid.org/http://orcid.org/0000-0003-1409-3562
https://orcid.org/http://orcid.org/0000-0001-6802-033X
Computer vision (CV) has been widely applied in many fields, including construction safety. As a factor of construction safety, ergonomic risk is receiving more and more attention. Although there are a few ergonomic risk assessment frameworks based on CV, the current mainstream CV-based ergonomic risk assessments cannot monitor and estimate the ergonomic risk of the human using all the necessary ergonomic risk information, such as the wrist angle. Most existing methods are based on detecting key points of the human body in images and give the two-dimensional (2D) pixel coordinates of the key points instead of the actual three-dimensional (3D) coordinates. Discrepancies exist between the skeletal angles calculated based on 2D and exact 3D coordinates. Rapid Entire Body Analysis (REBA) is one of the most common risk assessment tools researchers use, requiring accurate 3D joint angles. Therefore, this paper presents a comprehensive ergonomic risk assessment method based on CV and REBA. In addition to capturing coordinates throughout the body, it can estimate wrist angles in real time; thus, this method addresses the research gap. The proposed method utilizes CV to infer ten joints and six key points (e.g., head, thorax, etc.) throughout the body. With the estimated joints and key points, it can calculate six angles between different body segments and the twists of five body joints for REBA calculations. Data were collected in a laboratory with a retro-reflective marker-based motion capture system to simulate the operations of construction workers. The data were used to evaluate the accuracy of the proposed method. In total, the evaluation data consists of 80,300 annotated images. Results show that the proposed framework, in conjunction with the deep neural network models we have trained, can accurately estimate REBA-related key points of construction workers; therefore, the accurate ergonomics risk score for REBA can be obtained. It is worth noting that the proposed method can also be applied to workers in other fields beyond construction sites.
Real-Time Ergonomic Risk Assessment Approach for Construction Workers Based on Computer Vision
https://orcid.org/http://orcid.org/0000-0003-4367-4219
https://orcid.org/http://orcid.org/0000-0003-1409-3562
https://orcid.org/http://orcid.org/0000-0001-6802-033X
Computer vision (CV) has been widely applied in many fields, including construction safety. As a factor of construction safety, ergonomic risk is receiving more and more attention. Although there are a few ergonomic risk assessment frameworks based on CV, the current mainstream CV-based ergonomic risk assessments cannot monitor and estimate the ergonomic risk of the human using all the necessary ergonomic risk information, such as the wrist angle. Most existing methods are based on detecting key points of the human body in images and give the two-dimensional (2D) pixel coordinates of the key points instead of the actual three-dimensional (3D) coordinates. Discrepancies exist between the skeletal angles calculated based on 2D and exact 3D coordinates. Rapid Entire Body Analysis (REBA) is one of the most common risk assessment tools researchers use, requiring accurate 3D joint angles. Therefore, this paper presents a comprehensive ergonomic risk assessment method based on CV and REBA. In addition to capturing coordinates throughout the body, it can estimate wrist angles in real time; thus, this method addresses the research gap. The proposed method utilizes CV to infer ten joints and six key points (e.g., head, thorax, etc.) throughout the body. With the estimated joints and key points, it can calculate six angles between different body segments and the twists of five body joints for REBA calculations. Data were collected in a laboratory with a retro-reflective marker-based motion capture system to simulate the operations of construction workers. The data were used to evaluate the accuracy of the proposed method. In total, the evaluation data consists of 80,300 annotated images. Results show that the proposed framework, in conjunction with the deep neural network models we have trained, can accurately estimate REBA-related key points of construction workers; therefore, the accurate ergonomics risk score for REBA can be obtained. It is worth noting that the proposed method can also be applied to workers in other fields beyond construction sites.
Real-Time Ergonomic Risk Assessment Approach for Construction Workers Based on Computer Vision
https://orcid.org/http://orcid.org/0000-0003-4367-4219
https://orcid.org/http://orcid.org/0000-0003-1409-3562
https://orcid.org/http://orcid.org/0000-0001-6802-033X
Computer vision (CV) has been widely applied in many fields, including construction safety. As a factor of construction safety, ergonomic risk is receiving more and more attention. Although there are a few ergonomic risk assessment frameworks based on CV, the current mainstream CV-based ergonomic risk assessments cannot monitor and estimate the ergonomic risk of the human using all the necessary ergonomic risk information, such as the wrist angle. Most existing methods are based on detecting key points of the human body in images and give the two-dimensional (2D) pixel coordinates of the key points instead of the actual three-dimensional (3D) coordinates. Discrepancies exist between the skeletal angles calculated based on 2D and exact 3D coordinates. Rapid Entire Body Analysis (REBA) is one of the most common risk assessment tools researchers use, requiring accurate 3D joint angles. Therefore, this paper presents a comprehensive ergonomic risk assessment method based on CV and REBA. In addition to capturing coordinates throughout the body, it can estimate wrist angles in real time; thus, this method addresses the research gap. The proposed method utilizes CV to infer ten joints and six key points (e.g., head, thorax, etc.) throughout the body. With the estimated joints and key points, it can calculate six angles between different body segments and the twists of five body joints for REBA calculations. Data were collected in a laboratory with a retro-reflective marker-based motion capture system to simulate the operations of construction workers. The data were used to evaluate the accuracy of the proposed method. In total, the evaluation data consists of 80,300 annotated images. Results show that the proposed framework, in conjunction with the deep neural network models we have trained, can accurately estimate REBA-related key points of construction workers; therefore, the accurate ergonomics risk score for REBA can be obtained. It is worth noting that the proposed method can also be applied to workers in other fields beyond construction sites.
Real-Time Ergonomic Risk Assessment Approach for Construction Workers Based on Computer Vision
Lecture Notes in Civil Engineering
Desjardins, Serge (editor) / Poitras, Gérard J. (editor) / Nik-Bakht, Mazdak (editor) / Fan, Chao (author) / Mei, Qipei (author) / Li, Xinming (author)
Canadian Society of Civil Engineering Annual Conference ; 2023 ; Moncton, NB, Canada
Proceedings of the Canadian Society for Civil Engineering Annual Conference 2023, Volume 5 ; Chapter: 9 ; 113-127
2024-12-18
15 pages
Article/Chapter (Book)
Electronic Resource
English