Distinctive 3D local deep descriptors

1 Sep 2020  ·  Fabio Poiesi, Davide Boscaini ·

We present a simple but yet effective method for learning distinctive 3D local deep descriptors (DIPs) that can be used to register point clouds without requiring an initial alignment. Point cloud patches are extracted, canonicalised with respect to their estimated local reference frame and encoded into rotation-invariant compact descriptors by a PointNet-based deep neural network. DIPs can effectively generalise across different sensor modalities because they are learnt end-to-end from locally and randomly sampled points. Because DIPs encode only local geometric information, they are robust to clutter, occlusions and missing regions. We evaluate and compare DIPs against alternative hand-crafted and deep descriptors on several indoor and outdoor datasets consisting of point clouds reconstructed using different sensors. Results show that DIPs (i) achieve comparable results to the state-of-the-art on RGB-D indoor scenes (3DMatch dataset), (ii) outperform state-of-the-art by a large margin on laser-scanner outdoor scenes (ETH dataset), and (iii) generalise to indoor scenes reconstructed with the Visual-SLAM system of Android ARCore. Source code: https://github.com/fabiopoiesi/dip.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Point Cloud Registration 3DMatch Benchmark DIP Feature Matching Recall 94.8 # 9
Point Cloud Registration ETH (trained on 3DMatch) DIP Feature Matching Recall 0.928 # 2
Recall (30cm, 5 degrees) 62.41 # 10
Point Cloud Registration FPv1 DIP Recall (3cm, 10 degrees) 54.81 # 3
RRE (degrees) 4.058 # 3
RTE (cm) 2.052 # 1
Point Cloud Registration KITTI DIP Success Rate 97.30 # 4
Point Cloud Registration KITTI (trained on 3DMatch) DIP Success Rate 93.51 # 6

Methods


No methods listed for this paper. Add relevant methods here