Synthetic Dataset Generation for Partially Observed Indoor Objects

TL;DR

This paper introduces a Unity-based virtual scanning framework for generating realistic synthetic indoor 3D scan datasets, facilitating training and evaluation of scene reconstruction methods without costly real-world data collection.

Contribution

The authors present a novel virtual scanning system that simulates real-world sensor behavior and integrates with procedural scene generation to create diverse, annotated synthetic datasets.

Findings

Created the V-Scan dataset with realistic partial and complete 3D indoor scans.

Simulated sensor noise and occlusion effects for realism.

Enabled scalable dataset generation for training learning-based models.

Abstract

Learning-based methods for 3D scene reconstruction and object completion require large datasets containing partial scans paired with complete ground-truth geometry. However, acquiring such datasets using real-world scanning systems is costly and time-consuming, particularly when accurate ground truth for occluded regions is required. In this work, we present a virtual scanning framework implemented in Unity for generating realistic synthetic 3D scan datasets. The proposed system simulates the behaviour of real-world scanners using configurable parameters such as scan resolution, measurement range, and distance-dependent noise. Instead of directly sampling mesh surfaces, the framework performs ray-based scanning from virtual viewpoints, enabling realistic modelling of sensor visibility and occlusion effects. In addition, panoramic images captured at the scanner location are used to assign colours to the resulting point clouds. To support scalable dataset creation, the scanner is integrated with a procedural indoor scene generation pipeline that automatically produces diverse room layouts and furniture arrangements. Using this system, we introduce the \textit{V-Scan} dataset, which contains synthetic indoor scans together with object-level partial point clouds, voxel-based occlusion grids, and complete ground-truth geometry. The resulting dataset provides valuable supervision for training and evaluating learning-based methods for scene reconstruction and object completion.