PREDATOR: Registration of 3D Point Clouds with Low Overlap

Shengyu Huang, Zan Gojcic, Mikhail Usvyatsov, Andreas Wieser, Konrad Schindler
11/25/2020
cs.CVeess.IV

Official code: prs-eth/OverlapPredator (606 stars, MIT, last commit 2025-02-20).

Abstract

We introduce PREDATOR, a model for pairwise point-cloud registration with deep attention to the overlap region. Different from previous work, our model is specifically designed to handle (also) point-cloud pairs with low overlap. Its key novelty is an overlap-attention block for early information exchange between the latent encodings of the two point clouds. In this way the subsequent decoding of the latent representations into per-point features is conditioned on the respective other point cloud, and thus can predict which points are not only salient, but also lie in the overlap region between the two point clouds. The ability to focus on points that are relevant for matching greatly improves performance: PREDATOR raises the rate of successful registrations by more than 20% in the low-overlap scenario, and also sets a new state of the art for the 3DMatch benchmark with 89% registration recall.

View on arXivView PDF

Code Implementations(1)

prs-eth/OverlapPredatorOfficial100%
MIT

[CVPR 2021, Oral] PREDATOR: Registration of 3D Point Clouds with Low Overlap.

60689C++, ShellNov 24, 20201 years agoMIT
3dvisionattention-mechanismpoint-cloudregistrationtransformer

Similar Papers

PEAL: Prior-embedded Explicit Attention Learning for Low-overlap Point Cloud Registration

90% match1 repo582023

A Unified BEV Model for Joint Learning of 3D Local Features and Overlap Estimation

89% match2023

Overlap Bias Matching is Necessary for Point Cloud Registration

88% match2023

RoCNet: 3D Robust Registration of Point-Clouds using Deep Learning

87% match2023

Fully-Geometric Cross-Attention for Point Cloud Registration

87% match2025

LAHNet: Local Attentive Hashing Network for Point Cloud Registration

87% match2025

Cite this paper

@article{huang2020predator,
  title  = {PREDATOR: Registration of 3D Point Clouds with Low Overlap},
  author = {Shengyu Huang and Zan Gojcic and Mikhail Usvyatsov and Andreas Wieser and Konrad Schindler},
  year   = {2020},
  eprint = {2011.13005},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/2011.13005}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2011.13005.svg)](https://codeofpaper.com/paper/2011.13005)

Discussion