Point-GN: A Non-Parametric Network Using Gaussian Positional Encoding for Point Cloud Classification
Published in 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2025
This paper proposes Point-GN, a novel non-parametric network that uses Gaussian Positional Encoding to enable accurate and efficient 3D point cloud classification. By avoiding trainable parameters and relying on geometric operations like FPS and k-NN, Point-GN delivers competitive accuracy while being highly efficient and scalable—ideal for real-time or resource-constrained applications.