Modeling Point Clouds With Self-Attention and Gumbel Subset Sampling
Authors:
- 1 June 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 3318-3327
Abstract
Geometric deep learning is increasingly important thanks to the popularity of 3D sensors. Inspired by the recent advances in NLP domain, the self-attention transformer is introduced to consume the point clouds. We develop Point Attention Transformers (PATs), using a parameter-efficient Group Shuffle Attention (GSA) to replace the costly Multi-Head Attention. We demonstrate its ability to process size-varying inputs, and prove its permutation equivariance. Besides, prior work uses heuristics dependence on the input data (e.g., Furthest Point Sampling) to hierarchically select subsets of input points. Thereby, we for the first time propose an end-to-end learnable and task-agnostic sampling operation, named Gumbel Subset Sampling (GSS), to select a representative subset of input points. Equipped with Gumbel-Softmax, it produces a "soft" continuous subset in training phase, and a "hard" discrete subset in test phase. By selecting representative subsets in a hierarchical fashion, the networks learn a stronger representation of the input sets with lower computation cost. Experiments on classification and segmentation benchmarks show the effectiveness and efficiency of our methods. Furthermore, we propose a novel application, to process event camera stream as point clouds, and achieve a state-of-the-art performance on DVS128 Gesture Dataset.Keywords
All Related Versions
This publication has 25 references indexed in Scilit:
- Fine-Grained Recognition via Attribute-Guided Attentive Feature Aggregation2017
- Predicting Human Interaction via Relative Attention Model2017
- O-CNNACM Transactions on Graphics, 2017
- Geometric Deep Learning: Going beyond Euclidean dataIEEE Signal Processing Magazine, 2017
- Volumetric and Multi-view CNNs for Object Classification on 3D Data2016
- Deep Residual Learning for Image Recognition2016
- 3D Semantic Parsing of Large-Scale Indoor Spaces2016
- Multi-view Convolutional Neural Networks for 3D Shape Recognition2015
- Effective Approaches to Attention-based Neural Machine Translation2015
- A 128$\times$128 120 dB 15 $\mu$s Latency Asynchronous Temporal Contrast Vision SensorIEEE Journal of Solid-State Circuits, 2008