Point-Voxel CNN for Efficient 3D Deep Learning

Authors: Zhijian Liu, Haotian Tang, Yujun Lin, Song Han

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Evaluated on semantic and part segmentation datasets, it achieves a much higher accuracy than the voxel-based baseline with 10 GPU memory reduction; it also outperforms the state-of-the-art point-based models with 7 measured speedup on average.
Researcher Affiliation Academia Zhijian Liu MIT Haotian Tang Shanghai Jiao Tong University Yujun Lin MIT Song Han MIT
Pseudocode No No explicit pseudocode or algorithm block found.
Open Source Code No No explicit statement or link indicating the release of the authors' source code.
Open Datasets Yes Shape Net Parts [3]
Dataset Splits Yes We follow Qi et al. [29] to construct the val set from the training set so that no instances in the val set belong to the same video clip of any training instance. The size of val set is 3769, leaving the other 3711 samples for training.
Hardware Specification Yes We report the measured latency and GPU memory consumption on a single GTX 1080Ti GPU.
Software Dependencies No The paper does not specify version numbers for any software dependencies.
Experiment Setup Yes We ensure the input data to have the same size with 2048 points and batch size of 8. We measure the latency and memory consumption with 32768 points per batch at test time on a single GTX 1080Ti GPU.