What question did this study set out to answer?

To develop an automated instance segmentation framework for indoor 3D point cloud datasets combining unsupervised geometric segmentation with a self-learning ontology.

August 12, 2020The international archives of the photogrammetry, remote sensing and spatial information sciences/International archives of the photogrammetry, remote sensing and spatial information sciencesOpen Access

Self-Learning Ontology for Instance Segmentation of 3d Indoor Point Cloud

View Full Paper

Ask AI

Bookmark

Authors

FPFlorent PouxUniversity of Liège JPJean-Jacques PoncianoUniversity of Applied Sciences Mainz

Discussion

Member takes

Implication

Benchmarking study demonstrates 99.99% precision for planar classes in indoor 3D point clouds, highlighting automated ontology-driven segmentation over deep learning.

Key Points

To develop an automated instance segmentation framework for indoor 3D point cloud datasets combining unsupervised geometric segmentation with a self-learning ontology.
Extracted raw 3D coordinate shape features and voxel topological relationships to generate structural connectivity descriptions.
Applied planar unsupervised clustering followed by semantic classification using OWL2 and SPARQL queries within an iteratively refined ontology.
Benchmarked segmentation performance against state-of-the-art deep learning methods using the Stanford Large-Scale 3D Indoor Spaces (S3DIS) dataset.
Demonstrated fully autonomous, unsupervised instance segmentation without reliance on pre-annotated training data.
Achieved 99.99% precision for planar-dominant structural classes, outperforming existing state-of-the-art deep learning architectures.

Cite This Study

Poux et al. (2020) studied this question.

synapsesocial.com/papers/6a10dda5ba20d9a181ee646f https://doi.org/10.5194/isprs-archives-xliii-b2-2020-309-2020

View Full Paper

Ask AI

Bookmark

Implication

Benchmarking study demonstrates 99.99% precision for planar classes in indoor 3D point clouds, highlighting automated ontology-driven segmentation over deep learning.

Self-Learning Ontology for Instance Segmentation of 3d Indoor Point Cloud

Discussion

Member takes

Implication

Key Points

Cite This Study

Self-Learning Ontology for Instance Segmentation of 3d Indoor Point Cloud

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider