MinkLoc3D: Point Cloud Based Large-Scale Place Recognition

9 Nov 2020  ·  Jacek Komorowski ·

The paper presents a learning-based method for computing a discriminative 3D point cloud descriptor for place recognition purposes. Existing methods, such as PointNetVLAD, are based on unordered point cloud representation. They use PointNet as the first processing step to extract local features, which are later aggregated into a global descriptor. The PointNet architecture is not well suited to capture local geometric structures. Thus, state-of-the-art methods enhance vanilla PointNet architecture by adding different mechanism to capture local contextual information, such as graph convolutional networks or using hand-crafted features. We present an alternative approach, dubbed MinkLoc3D, to compute a discriminative 3D point cloud descriptor, based on a sparse voxelized point cloud representation and sparse 3D convolutions. The proposed method has a simple and efficient architecture. Evaluation on standard benchmarks proves that MinkLoc3D outperforms current state-of-the-art. Our code is publicly available on the project website: https://github.com/jac99/MinkLoc3D

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
3D Place Recognition CS-Campus3D Minkloc3D AR@1% 79.1 # 2
AR@1 69.38 # 2
AR@1% cross-source 85.22 # 2
AR@1 cross-source 55.36 # 2
Point Cloud Retrieval Oxford RobotCar (LiDAR 4096 points) MinkLoc3D (refined) recall@top1% 98.5 # 4
recall@top1 94.80 # 5
Point Cloud Retrieval Oxford RobotCar (LiDAR 4096 points) MinkLoc3D (baseline) recall@top1% 97.9 # 11

Methods