.png){kind=icon}
.png){kind=icon}
Return
Automatic defect localization method based on BIM projected image
.png){kind=icon}
PRE
AI
DOI：10.1016/j.aei.2025.103874.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Deficiencies are prevalent in building construction. Prompt inspection and precise identification of problems are essential to guarantee building quality. Traditional defect approaches depend on manual documentation following inspections or utilize sensors like Global Positioning System (GPS) and inertial measurement units. These procedures are inefficient, expensive, constrained by equipment conditions, and incapable of identifying flaws in specific construction components. To tackle the aforementioned issues, this study offers a new defect localization framework utilizing BIM-projected pictures to attain accurate defect localization of individual components from a single image. The framework relies on a repository of Building Information Modeling (BIM) projected images with established positional projections. By effectively retrieving and matching features of live images collected on-site with the library, the camera position of the background scene in the live images is precisely estimated. Consequently, based on this accurate position and the 2D information of the defects in the images, the defects are accurately mapped to the corresponding BIM components. The technology facilitates automated, high-precision localization of problems at the component level without requiring further financial expenditure. Experimental results validate that the suggested method excels in camera position estimation. The precision of its validation on defect-free photos is exceptional, and this benefit is effectively transferred to the validation of real images, demonstrating excellent accuracy and practical utility.
Journal
IF:
9.9
Papers:
4.0K
Citations:
1.7W