Reconstructing the MI. Building in a Day

We offer a master (or guided research) topic in the field of computer vision and robotics, which aims to reconstruct large-scale indoor environments efficiently.

Abstract

Planar regions including floors, walls, ceilings and large surfaces of furniture are common elements in indoor scenes, and even main features of those environments. How to deal with those regions affects the scale we can reconstruct with limited computation. In this project, we make use of a consumer RGB-D camera and a CPU to reconstruct the main structure of the MI. building, which contains a few room scenes, corridors and stairs.

Literature

• [1]. Wang et al. Plane-based optimization of geometry and texture for rgb-d reconstruction of indoor scenes. 3DV2018.
• [2]. Fu et al. Joint Texture and Geometry Optimization for RGB-D Reconstruction. CVPR2020
• [3]. Shopes et al. SurfelMeshing^?: Online Surfel-Based Mesh Reconstruction. PAMI 2019.
• [4]. Dai et al. BundleFusion^?: Real-time Globally Consistent 3D Reconstruction using On-the-fly Surface Re-integration. TOG2017
• [5]. Han et al. FlashFusion^?: Real-time Globally Consistent Dense 3D Reconstruction using CPU Computing. RSS2018
• [6]. Choi et al. Robust Reconstruction of Indoor Scenes. CVPR2015
• [7]. Wang et al. Real-time Scalable Dense Surfel Mapping. ICRA2019
• [8]. Whelan et al. ElasticFusion^?: Dense SLAM Without A Pose Graph. RSS2015
• [9]. Campos et al. ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual-Inertial and Multi-Map SLAM. Arxiv2020

Requirements

• A strong background in SLAM/SfM. It would be better if you modified ORB-SLAM(V1,2,3) or other SLAM systems before.
• Good C++ skill

If you are interested in any of these topics, please contact us via e-mail.

Yanyan Li

Federico Tombari