PCFootprint: A Large-Scale Dataset and Benchmark for Vectorized Building Footprint Extraction from Aerial LiDAR Point Clouds

2026-06-18 · Source: Computer Vision and Pattern Recognition · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Computer Vision · Depth: Expert, quick

Summary

PCFootprint introduces the first large-scale public dataset for vectorized building footprint extraction from airborne laser scanning point clouds. This dataset addresses inherent limitations of image-based methods, such as occlusions, perspective distortions, and lack of explicit elevation information. Comprising 33,000 tiles derived from the Estonian Land and Spatial Development Board, PCFootprint covers diverse urban and rural landscapes, with each tile spanning 128 x 128 m and featuring systematically aligned vectorized footprints. It includes a 3,000-tile cross-domain test set to evaluate generalization across geographic regions. Benchmarking mainstream methods on PCFootprint reveals significant challenges, including high intra-class variance, data imbalance, and noise in complex geospatial environments. The dataset is publicly available on Hugging Face.

Key takeaway

For computer vision engineers developing building footprint extraction models, traditional image-based methods face inherent limitations like occlusions and lack of elevation. You should consider integrating LiDAR point cloud data, as the new PCFootprint dataset provides a large-scale, diverse resource to train and benchmark models, improving robustness across varied urban and rural environments. Utilize this dataset to overcome current challenges and advance urban scene understanding.

Key insights

PCFootprint is the first large-scale LiDAR dataset for vectorized building footprint extraction, addressing optical imagery limitations.

Principles

• LiDAR overcomes optical imagery limits.
• Diverse datasets improve generalization.
• Benchmarking reveals extraction challenges.

Method

The article describes creating PCFootprint from 33,000 Estonian LiDAR tiles, each 128x128m, with aligned vectorized footprints, including a 3,000-tile cross-domain test set.

In practice

• Use PCFootprint for building modeling.
• Evaluate methods on diverse landscapes.
• Address data imbalance in LiDAR.

Topics

• Building Footprint Extraction
• LiDAR Point Clouds
• Aerial Laser Scanning
• Geospatial Analysis
• Urban Scene Understanding
• Dataset Benchmark

Best for: AI Scientist, Computer Vision Engineer, Research Scientist

Related on AIssential

• Building Extraction from Remote Sensing Imagery under Hazy and Low-light Conditions: Benchmark and Baseline — A new benchmark and network address building extraction challenges in hazy and low-light remote sensing imagery.
• AI Maps 13 Million Buildings in One of the World’s Most Remote Regions — AI-powered satellite imagery can accurately map millions of buildings in remote, challenging terrains.
• Integrating national forest inventory, airborne lidar, and satellite imagery for wall-to-wall mapping of forest structure with computer vision — Integrating lidar and satellite data with computer vision improves large-scale forest attribute mapping accuracy and consistency.
• OmniLiDAR: A Unified Diffusion Framework for Multi-Domain 3D LiDAR Generation — OmniLiDAR unifies 3D LiDAR generation across diverse sensing domains using a text-conditioned diffusion framework.
• Monocular Building Height Estimation from PhiSat-2 Imagery: Dataset and Method — PhiSat-2 imagery, combined with a novel two-stream ordinal network, significantly improves monocular building height estimation.
• Plan2Map: A Multimodal Benchmark for Document-Grounded Geospatial Boundary Reconstruction from Planning Records — Plan2Map enables multimodal geospatial boundary reconstruction from diverse planning documents via structured processing.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computer Vision and Pattern Recognition.