One Shot Learning for Edge Detection on Point Clouds

2026-04-24

AI Summary

This work addresses the significant performance degradation of existing point cloud edge detection methods when applied across different scanning devices, primarily due to their inability to adapt to sensor-specific sampling error distributions. To overcome this limitation, we propose the first single-sample learning framework tailored for point cloud edge detection, which adapts to the target data distribution using only a single example—eliminating the need for large-scale multi-source training data. Our approach integrates a filtered k-nearest neighbor (KNN)-based surface patch representation, a radial basis function surface descriptor (RBF_DoS), and a lightweight network, OSFENet, to effectively capture local geometric structures. Experiments demonstrate that our method outperforms seven baselines on the ABC dataset and exhibits strong generalization and practical utility across diverse real-world indoor and outdoor scenes, including S3DIS, Semantic3D, and UrbanBIS.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: 3D Computer Vision
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Each scanner possesses its unique characteristics and exhibits its distinct sampling error distribution. Training a network on a dataset that includes data collected from different scanners is less effective than training it on data specific to a single scanner. Therefore, we present a novel one-shot learning method allowing for edge extraction on point clouds, by learning the specific data distribution of the target point cloud, and thus achieve superior results compared to networks that were trained on general data distributions. More specifically, we present how to train a lightweight network named OSFENet (One-Shot edge Feature Extraction Network), by designing a filtered-KNN-based surface patch representation that supports a one-shot learning framework. Additionally, we introduce an RBF_DoS module, which integrates Radial Basis Function-based Descriptor of the Surface patch, highly beneficial for the edge extraction on point clouds. The advantage of the proposed OSFENet is demonstrated through comparative analyses against 7 baselines on the ABC dataset, and its practical utility is validated by results across diverse real-scanned datasets, including indoor scenes like S3DIS dataset, and outdoor scenes such as the Semantic3D dataset and UrbanBIS dataset.

Problem

• One-shot learning
• Edge detection
• Point clouds
• Scanner-specific distribution
• 3D data

Innovation

• one-shot learning
• point cloud edge detection
• filtered-KNN
• RBF_DoS
• OSFENet