Deep Models with Fusion Strategies for MVP Point Cloud Registration
Not on the map. Why?
Abstract (arXiv)
The main goal of point cloud registration in Multi-View Partial (MVP) Challenge 2021 is to estimate a rigid transformation to align a point cloud pair. The pairs in this competition have the characteristics of low overlap, non-uniform density, unrestricted rotations and ambiguity, which pose a huge challenge to the registration task. In this report, we introduce our solution to the registration task, which fuses two deep learning models: ROPNet and PREDATOR, with customized ensemble strategies. Finally, we achieved the second place in the registration track with 2.96546, 0.02632 and 0.07808 under the the metrics of RotError, TransError and MSE, respectively.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 9 of its 13 references that are in the map. First posted on arXiv 18 Oct 2021.
Builds on
9 of 13 references.
References · 9
- 2020PCT: Point cloud transformercited by 732
- 2018Open3D: A Modern Library for 3D Data Processingcited by 721
- 20163DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructionscited by 336
- 2019Deep Closest Point: Learning Representations for Point Cloud Registrationcited by 321
- 2020PREDATOR: Registration of 3D Point Clouds with Low Overlapcited by 290
- 2020RPM-Net: Robust Point Matching using Learned Featurescited by 182
- 2021Variational Relational Point Completion Networkcited by 104
- 2020DeepGMR: Learning Latent Gaussian Mixture Models for Registrationcited by 91
- 2020Correspondence Matrices are Underratedcited by 5
Built on by
Not cited yet.