Pointwise Rotation-Invariant Network with Adaptive Sampling and 3D Spherical Voxel Convolution

Authors: Yang You, Yujing Lou, Qi Liu, Yu-Wing Tai, Lizhuang Ma, Cewu Lu, Weiming Wang12717-12724

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We experimentally compare PRIN with various state-of-the-art approaches on the benchmark dataset: Shape Net part dataset (Yi et al. 2016) and Model Net40 (Wu et al. 2015).In this section, we arrange comprehensive expreriments to evaluate PRIN for point cloud analysis on different tasks.
Researcher Affiliation Collaboration 1Shanghai Jiao Tong University, China 2Tencent, China
Pseudocode No The paper describes its methods in prose and with diagrams, but does not include any explicit pseudocode or algorithm blocks.
Open Source Code Yes The code to reproduce our results is available online.1
Open Datasets Yes Shape Net part dataset (Yi et al. 2016) and Model Net40 (Wu et al. 2015)
Dataset Splits Yes We follow the data split in Point Net (Charles et al. 2017).
Hardware Specification Yes PRIN is implemented in Python using Py Torch on one NVIDIA TITAN Xp.
Software Dependencies No The paper states 'PRIN is implemented in Python using Py Torch', but does not provide specific version numbers for Python or PyTorch.
Experiment Setup Yes In all of our experiments, Adam optimization algorithm is employed for training, with a batch size of 16. The learning rate begins with 0.01 and decays with a rate of 0.5 every 5 epochs. During training we set the input bandwidth on S2 to 32 and resolution on H to 64. The threshold filter width δ = 1/32.