ColabSfM: Collaborative Structure-from-Motion by Point Cloud Registration
This work addresses the lack of scalable methods and datasets for registering SfM reconstructions, enabling collaborative 3D mapping, though it is incremental as it builds on existing registration techniques.
The paper tackles the challenge of registering distributed Structure-from-Motion (SfM) reconstructions for collaborative sharing by proposing a scalable point cloud registration task, a dataset generation pipeline, and a neural refiner called RefineRoITr, which significantly improves registration accuracy, enabling ColabSfM with demonstrated performance gains.
Structure-from-Motion (SfM) is the task of estimating 3D structure and camera poses from images. We define Collaborative SfM (ColabSfM) as sharing distributed SfM reconstructions. Sharing maps requires estimating a joint reference frame, which is typically referred to as registration. However, there is a lack of scalable methods and training datasets for registering SfM reconstructions. In this paper, we tackle this challenge by proposing the scalable task of point cloud registration for SfM reconstructions. We find that current registration methods cannot register SfM point clouds when trained on existing datasets. To this end, we propose a SfM registration dataset generation pipeline, leveraging partial reconstructions from synthetically generated camera trajectories for each scene. Finally, we propose a simple but impactful neural refiner on top of the SotA registration method RoITr that yields significant improvements, which we call RefineRoITr. Our extensive experimental evaluation shows that our proposed pipeline and model enables ColabSfM. Code is available at https://github.com/EricssonResearch/ColabSfM