BS3D: Building-scale 3D Reconstruction from RGB-D Images

TL;DR

This paper introduces BS3D, a practical framework for building-scale 3D reconstruction using consumer RGB-D cameras, enabling crowd-sourcing and improved data for SLAM and depth estimation tasks.

Contribution

It presents a novel, accessible system for large-scale 3D data acquisition that leverages raw depth maps and infrared data, filling gaps in existing datasets.

Findings

Enhanced 3D reconstructions from raw depth data.

Improved monocular depth estimation models trained on BS3D.

Benchmarking of visual-inertial odometry with infrared images.

Abstract

Various datasets have been proposed for simultaneous localization and mapping (SLAM) and related problems. Existing datasets often include small environments, have incomplete ground truth, or lack important sensor data, such as depth and infrared images. We propose an easy-to-use framework for acquiring building-scale 3D reconstruction using a consumer depth camera. Unlike complex and expensive acquisition setups, our system enables crowd-sourcing, which can greatly benefit data-hungry algorithms. Compared to similar systems, we utilize raw depth maps for odometry computation and loop closure refinement which results in better reconstructions. We acquire a building-scale 3D dataset (BS3D) and demonstrate its value by training an improved monocular depth estimation model. As a unique experiment, we benchmark visual-inertial odometry methods using both color and active infrared images.