A virtual construction vehicles and workers dataset with three-dimensional annotations

Most current artificial intelligence (AI) studies based on computer vision (CV) in construction are almost based on pure two-dimensional (2D) data, limiting their effectiveness to scenarios that only require 2D information. However, advanced AI systems often require the ability to perceive three-dimensional (3D) spatial information, which constrain their application scope in construction. To address this limitation, this study presents a Virtual Construction Vehicles and Workers Dataset with three-dimensional Annotations (VCVW-3D). The dataset covers 15 construction scenes and includes ten categories of construction vehicles and workers. The VCVW-3D dataset is characterized by its multi-scene, multi-category, multi-randomness, multi-viewpoint, multi-annotation, and binocular vision features. We trained and evaluated several 2D and monocular 3D object detection models, such as You Only Look Once (YOLO) and Fully Convolutional One-Stage Monocular 3D Object Detection (FCOS3D), using the VCVW-3D to establish benchmarks. The VCVW-3D dataset aims to promote the development of 3D computer vision in the construction industry by reducing the costs associated with data acquisition, prototype development, and exploration of space-awareness applications. © 2024 Elsevier Ltd. All rights reserved.