Defect Detection and Quality Control in 3D-Printed Construction Elements Using High-Resolution 3D Scanning and Deep Learning Models

3D printing is transforming the construction industry, particularly in modular housing, by enabling customizable, efficient, and cost-effective production. However, maintaining the quality of 3D-printed concrete walls remains a significant challenge due to common defects such as layer height inconsistencies, voids, clumps, and surface collapses, which compromise structural integrity and appearance. This study investigates advanced methods for detecting and classifying such defects using high-resolution 3D scanning integrated with deep learning models. By comparing scanned structures with original design files, the system enhances quality control, reduces material waste, and supports sustainability goals. The methodology introduces percentile-based thresholds for defect classification, adjusts mesh resolution for improved alignment, and employs a voxel-based 3D convolutional neural network (CNN) for accurate defect localization. This integrated approach increases the scalability, precision, and automation of quality assurance in additive manufacturing. The research provides a foundation for developing safer, more durable, and environmentally responsible 3D-printed modular homes.