EdgeFormer: local patch-based edge detection transformer on point clouds

2026-04-23

AI Summary

This work addresses the challenge of effectively detecting fine-grained edges in 3D point clouds—such as densely clustered edge points or regions with subtle surface gradients—by proposing a local patch-based point classification method. The approach constructs local neighborhood feature descriptors and integrates a two-stage learning architecture with a Transformer mechanism to explicitly model local surface correlations, thereby significantly enhancing the perception of intricate edge structures. Experimental results demonstrate that the proposed method consistently outperforms six state-of-the-art baselines across multiple benchmark datasets, achieving superior performance in both edge detection accuracy and preservation of geometric details.

Technology Category

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

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Edge points on 3D point clouds can clearly convey 3D geometry and surface characteristics, therefore, edge detection is widely used in many vision applications with high industrial and commercial demands. However, the fine-grained edge features are difficult to detect effectively as they are generally densely distributed or exhibit small-scale surface gradients. To address this issue, we present a learning-based edge detection network, named EdgeFormer, which mainly consists of two stages. Based on the observation that spatially neighboring points tend to exhibit high correlation, forming the local underlying surface, we convert the edge detection of the entire point cloud into a point classification based on local patches. Therefore, in the first stage, we construct local patch feature descriptors that describe the local neighborhood around each point. In the second stage, we classify each point by analyzing the local patch feature descriptors generated in the first stage. Due to the conversion of the point cloud into local patches, the proposed method can effectively extract the finer details. The experimental results show that our model demonstrates competitive performance compared to six baselines.

Problem

• edge detection
• 3D point clouds
• fine-grained features
• surface gradients
• local patches

Innovation

• EdgeFormer
• local patch
• edge detection
• point cloud
• transformer