Siddula, M; Dai, F; Ye, Y and Fan, J (2016) Classifying construction site photos for roof detection: A machine-learning method towards automated measurement of safety performance on roof sites. Construction Innovation, 16(3), pp. 368-389. ISSN 1471-4175
Abstract
Purpose Roofing is one of the most dangerous jobs in the construction industry. Due to factors such as lack of planning, training and use of precaution, roofing contractors and workers continuously violate the fall protection standards enforced by the US Occupational Safety and Health Administration. A preferable way to alleviate this situation is automating the process of non-compliance checking of safety standards through measurements conducted in site daily accumulated videos and photos. As a key component, the purpose of this paper is to devise a method to detect roofs in site images that is indispensable for such automation process. Design/methodology/approach This method represents roof objects through image segmentation and visual feature extraction. The visual features include colour, texture, compactness, contrast and the presence of roof corner. A classification algorithm is selected to use the derived representation for statistical learning and detection. Findings The experiments led to detection accuracy of 97.50 per cent, with over 15 per cent improvement in comparison to conventional classifiers, signifying the effectiveness of the proposed method. Research limitations/implications This study did not test on images of roofs in the following conditions: roofs initially built without apparent appearance (e.g. structural roof framing completed and undergoing the sheathing process) and flat, barrel and dome roofs. From a standpoint of construction safety, while the present work is vital, coupling with semantic representation and analysis is still needed to allow for risk analysis of fall violations on roof sites. Originality/value This study is the first to address roof detection in site images. Its findings provide a basis to enable semantic representation of roof site objects of interests (e.g. co-existence and correlation among roof site, roofer, guardrail and personal fall arrest system) that is needed to automate the non-compliance checking of safety standards on roof sites.
| Item Type: | Article |
|---|---|
| Uncontrolled Keywords: | construction management,construction safety,health and safety,machine learning,image-based methods,roofing industry |
| Index terms: | coupling, automation, roof, construction industry, roofer, occupational safety and health, effectiveness, presence, construction site, image-based method, framing, accuracy, experiment, compliance, health and safety, risk analysis, construction safety, safety standards, machine learning, violation, safety performance, roofing, methodology |
| Subjects: | work location, conceptual models, environmental science, artificial intelligence, data collection methods, practitioner, occupational health and safety management, systems engineering, environmental hazards, environmental health, regulatory law, industry analysis, performance management, research methods, design practice, professional development, health safety and environment, automation and robotics |
| Topics: | Quality Management, Legal Issues, Design Practice, Digital Applications, Site Management, Research Practice, Information Management, Engineering Principles, Health and Safety, Sustainability, Roles and Professions |
| Descriptive scope: | 4 PCTE |
N.B. Descriptive scope is a count of how many of the five facets of empirical research are indicated by the words used in title, abstract and keywords. It is not intended as a judgement on the research; merely a count of the kind of word we would expect to indicate Phenomenon, Concepts, Theoretical framing, Empirical techniques, Analytical techniques. If all five are present, then a code of “5 PCTEA” will indicate this. If you feel the coding for this record is questionable, we welcome discussion around the terms we matched or the way we categorized them. The facet you would expect may not be coded, or a facet may be coded inappropriately. This can also bear on a larger question, of which facets should be treated as defining in construction management research. Please get in touch, and we will look at it. More details here