Parameter is Not All You Need: Starting from Non-Parametric Networks for 3D Point Cloud Analysis
This work addresses 3D point cloud analysis for computer vision researchers by proposing a novel non-parametric approach that challenges the need for learnable parameters, offering high efficiency and complementary geometric insights.
The authors tackled 3D point cloud analysis by introducing Point-NN, a non-parametric network using only non-learnable components like FPS and k-NN, which surprisingly outperforms fully trained models on various tasks without parameters or training. They extended it to create efficient parametric networks and enhance existing models during inference, achieving strong performance on benchmarks.
We present a Non-parametric Network for 3D point cloud analysis, Point-NN, which consists of purely non-learnable components: farthest point sampling (FPS), k-nearest neighbors (k-NN), and pooling operations, with trigonometric functions. Surprisingly, it performs well on various 3D tasks, requiring no parameters or training, and even surpasses existing fully trained models. Starting from this basic non-parametric model, we propose two extensions. First, Point-NN can serve as a base architectural framework to construct Parametric Networks by simply inserting linear layers on top. Given the superior non-parametric foundation, the derived Point-PN exhibits a high performance-efficiency trade-off with only a few learnable parameters. Second, Point-NN can be regarded as a plug-and-play module for the already trained 3D models during inference. Point-NN captures the complementary geometric knowledge and enhances existing methods for different 3D benchmarks without re-training. We hope our work may cast a light on the community for understanding 3D point clouds with non-parametric methods. Code is available at https://github.com/ZrrSkywalker/Point-NN.