Structured Semantic 3D Reconstruction (S23DR) Challenge 2025 -- Winning solution

TL;DR

This paper introduces a novel 3D deep learning approach for reconstructing house roof wireframes from sparse point clouds, winning the S23DR Challenge 2025 with a high accuracy score.

Contribution

It presents a two-stage 3D deep learning method that directly predicts roof wireframes from point clouds, combining vertex classification and edge prediction.

Findings

Achieved a Hybrid Structure Score of 0.43 on the private leaderboard.

Demonstrated effectiveness of a 3D deep learning pipeline for semantic 3D reconstruction.

Utilized a combination of vertex candidate identification and edge prediction in 3D.

Abstract

This paper presents the winning solution for the S23DR Challenge 2025, which involves predicting a house's 3D roof wireframe from a sparse point cloud and semantic segmentations. Our method operates directly in 3D, first identifying vertex candidates from the COLMAP point cloud using Gestalt segmentations. We then employ two PointNet-like models: one to refine and classify these candidates by analyzing local cubic patches, and a second to predict edges by processing the cylindrical regions connecting vertex pairs. This two-stage, 3D deep learning approach achieved a winning Hybrid Structure Score (HSS) of 0.43 on the private leaderboard.