Research Status and Challenges of Intelligent Recognition of Prefabricated Construction Quality Based on BIM and Machine Vision
Keywords:
Digital Construction; BIM; Machine Vision; Prefabricated Construction; Quality Inspection; Defect DetectionAbstract
Prefabricated construction is widely regarded as a vital pathway to advance the industrialization, decarbonization and digital transformation of the construction sector. Compared with conventional cast-in-situ construction, prefabricated manufacturing transfers most production work to controlled factory workshops, yet on-site assembly quality remains a decisive factor governing structural safety, durability and seismic performance. Traditional quality inspection heavily relies on manual visual check and contact measurement, which suffers from low throughput, strong subjectivity, inconsistent recording and limited coverage of concealed defects. The fusion of Building Information Modeling (BIM) and machine vision offers a promising non-contact intelligent inspection paradigm: BIM provides standardized design benchmarks, dimensional tolerances and component spatial information, while machine vision captures site visual data, extracts defect features and quantifies assembly deviations. This paper systematically reviews recent advances in BIM-machine vision integrated inspection for prefabricated construction. It categorizes mainstream vision algorithms and compares their applicability to three typical defects: assembly joint misalignment, rebar layout irregularity and concrete cover thickness non-compliance. This study identifies core bottlenecks including poor scene generalization, virtual-physical domain mismatch, high computing overhead and lack of unified inspection standards. Differing from existing review papers that merely list algorithm achievements, this work proposes a lightweight decoupled inspection framework and highlights small-sample transfer learning as a feasible route for field deployment. The findings are expected to provide theoretical references and actionable directions for future research on digital construction quality control.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Wenjun Wang (Author)

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
