Point Cloud Oversegmentation with Graph-Structured Deep Metric Learning
| Autor: | Loic Landrieu, Mohamed Boussaha |
|---|---|
| Přispěvatelé: | Méthodes d'Analyses pour le Traitement d'Images et la Stéréorestitution (MATIS), Laboratoire des Sciences et Technologies de l'Information Géographique (LaSTIG), École nationale des sciences géographiques (ENSG), Institut National de l'Information Géographique et Forestière [IGN] (IGN)-Institut National de l'Information Géographique et Forestière [IGN] (IGN)-École nationale des sciences géographiques (ENSG), Institut National de l'Information Géographique et Forestière [IGN] (IGN)-Institut National de l'Information Géographique et Forestière [IGN] (IGN), Landrieu, Loic |
| Jazyk: | angličtina |
| Rok vydání: | 2019 |
| Předmět: |
FOS: Computer and information sciences
Computer Science - Machine Learning Theoretical computer science Computer science Computer Vision and Pattern Recognition (cs.CV) Computer Science - Computer Vision and Pattern Recognition [INFO.INFO-NE] Computer Science [cs]/Neural and Evolutionary Computing [cs.NE] 0211 other engineering and technologies Point cloud 02 engineering and technology [INFO.INFO-NE]Computer Science [cs]/Neural and Evolutionary Computing [cs.NE] Machine Learning (cs.LG) [INFO.INFO-CV] Computer Science [cs]/Computer Vision and Pattern Recognition [cs.CV] [STAT.ML]Statistics [stat]/Machine Learning [stat.ML] 0202 electrical engineering electronic engineering information engineering Segmentation 021101 geological & geomatics engineering Artificial neural network business.industry Deep learning Graph partition [INFO.INFO-CV]Computer Science [cs]/Computer Vision and Pattern Recognition [cs.CV] [STAT.ML] Statistics [stat]/Machine Learning [stat.ML] Graph (abstract data type) 020201 artificial intelligence & image processing Artificial intelligence business Feature learning |
| Zdroj: | CVPR CVPR, 2019, Long Beach, France HAL |
| Popis: | We propose a new supervized learning framework for oversegmenting 3D point clouds into superpoints. We cast this problem as learning deep embeddings of the local geometry and radiometry of 3D points, such that the border of objects presents high contrasts. The embeddings are computed using a lightweight neural network operating on the points' local neighborhood. Finally, we formulate point cloud oversegmentation as a graph partition problem with respect to the learned embeddings. This new approach allows us to set a new state-of-the-art in point cloud oversegmentation by a significant margin, on a dense indoor dataset (S3DIS) and a sparse outdoor one (vKITTI). Our best solution requires over five times fewer superpoints to reach similar performance than previously published methods on S3DIS. Furthermore, we show that our framework can be used to improve superpoint-based semantic segmentation algorithms, setting a new state-of-the-art for this task as well. CVPR2019 |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|