Abstract Deep learning (DL), a subset of artificial intelligence, has revolutionized data processing and decision-making across various domains, including construction. Point clouds, which represent spatial data in 3D space, have emerged as a critical format in construction for tasks such as site monitoring, structural analysis, and design validation. The integration of DL techniques with point cloud data offers transformative potential, enabling unprecedented accuracy and efficiency in complex construction processes. Despite its promise, the field remains underexplored, with limited comprehensive reviews addressing its applications and challenges. This paper consists in a concise review of current DL applications for point cloud data within the construction industry, systematically analysing their benefits, limitations, and prospects. In addition, it provides an overview of foundational DL algorithms, with a focus on their adaptation to point cloud data, and examines real-world and synthetic datasets that support these applications. Finally, strategic recommendations are suggested to enhance the advancement and adoption of DL-powered point cloud solutions, aiming to establish a roadmap for future advancements in this rapidly evolving domain.
Ungureanu et al. (Fri,) studied this question.