Englische Zusammenfassung
The project develops GF4Shape, combining geometric and learned features through a diffusion approach for point-cloud shape completion. Students will design a geometric-feature-based regularizer, implement and evaluate the method on indoor and urban outdoor environments, and compare it with M1 results. The planned work includes literature review, prototyping, dataset evaluation, report writing, and a final presentation.
Überblick
- Projektdetails
- GF4Shape: Geometric Features for Point Cloud Shape Completion
Textstelle · Seite 1
- “Project title: GF4Shape: Geometric Features for Point Cloud Shape Completion” Seite 1
- Organisation
- Im PDF nicht angegeben
- Projektformate
- Im PDF nicht angegeben
- Studienabschlüsse
- Im PDF nicht angegeben
- Projektziel
- Leverage geometric priors as a structural regularizer for shape completion, bridging the gap between accuracy and generalization in complex environments from aerial to terrestrial acquisition.
Textstelle · Seite 1
- “Our goal is to leverage geometric priors as structural regularizer for shape completion, aiming to bridge the gap between accuracy and generalization in complex environments from aerial to terrestrial acquisition.” Seite 1
Themen und Aufgaben
- Themen
- 3D point cloud shape completion
Textstelle · Seite 1
- “3D shape completion in point clouds remains fundamentally unsolved challenge.” Seite 1
- geometric features and priors
Textstelle · Seite 1
- “Our preliminary research indicates that integrating explicit geometric features (e.g., anisotropy, roughness, planarity) with learned features substantially improves the network’s performance for semantic segmentation accelerating convergence and enhancing robustness across datasets.” Seite 1
- learned features
Textstelle · Seite 1
- “Develop a method, GF4Shape, that fuses geometric features and learned features based on the diffusion approach.” Seite 1
- diffusion approach
Textstelle · Seite 1
- “Develop a method, GF4Shape, that fuses geometric features and learned features based on the diffusion approach.” Seite 1
- 3D point cloud shape completion
- Anwendungsbereiche
- indoor environments
Textstelle · Seite 1
- “Evaluate the GF4Shape shape completion method on indoor and urban outdoor environments.” Seite 1
- urban outdoor environments
Textstelle · Seite 1
- “Evaluate the GF4Shape shape completion method on indoor and urban outdoor environments.” Seite 1
- aerial and terrestrial acquisition
Textstelle · Seite 1
- “Our goal is to leverage geometric priors as structural regularizer for shape completion, aiming to bridge the gap between accuracy and generalization in complex environments from aerial to terrestrial acquisition.” Seite 1
- indoor environments
- Methoden und Werkzeuge
- geometric features
Textstelle · Seite 1
- “Develop a method, GF4Shape, that fuses geometric features and learned features based on the diffusion approach.” Seite 1
- learned features
Textstelle · Seite 1
- “Develop a method, GF4Shape, that fuses geometric features and learned features based on the diffusion approach.” Seite 1
- diffusion approach
Textstelle · Seite 1
- “Develop a method, GF4Shape, that fuses geometric features and learned features based on the diffusion approach.” Seite 1
- geometric-feature-based regularizer
Textstelle · Seite 1
- “Design geometric-feature-based regularizer for shape completion and compare to the M1 results.” Seite 1
- fitting datasets
Textstelle · Seite 2
- “Evaluation on fitting datasets” Seite 2
- geometric features
- Aufgaben
- Familiarisation with relevant papers and literature review
Textstelle · Seite 1
- “Familiarisation with relevant papers, literature review” Seite 1
- Propose several ideas for approaching the problem
Textstelle · Seite 1
- “Proposal of several ideas of how to approach the problem” Seite 1
- Prototype ideas
Textstelle · Seite 1
- “Prototyping of ideas” Seite 1
- Select the most promising path
Textstelle · Seite 2
- “Select the most promising path” Seite 2
- Achieve a basic implementation
Textstelle · Seite 2
- “Achieve a basic implementation” Seite 2
- Develop GF4Shape to fuse geometric and learned features
Textstelle · Seite 1
- “Develop a method, GF4Shape, that fuses geometric features and learned features based on the diffusion approach.” Seite 1
- Design a geometric-feature-based regularizer and compare it to M1 results
Textstelle · Seite 1
- “Design geometric-feature-based regularizer for shape completion and compare to the M1 results.” Seite 1
- Evaluate the shape completion method on indoor and urban outdoor environments
Textstelle · Seite 1
- “Evaluate the GF4Shape shape completion method on indoor and urban outdoor environments.” Seite 1
- Finalize the implementation
Textstelle · Seite 2
- “Finalizing of implementation” Seite 2
- Evaluate on fitting datasets
Textstelle · Seite 2
- “Evaluation on fitting datasets” Seite 2
- Write a report
Textstelle · Seite 2
- “Report writing” Seite 2
- Give a final presentation
Textstelle · Seite 2
- “Final presentation” Seite 2
- Familiarisation with relevant papers and literature review
- Art der Arbeit
- Literaturrecherche
Textstelle · Seite 1
- “Familiarisation with relevant papers, literature review” Seite 1
- Softwareentwicklung
Textstelle · Seite 2
- “Achieve a basic implementation” Seite 2
- Datenanalyse und ML
Textstelle · Seite 1
- “Evaluate the GF4Shape shape completion method on indoor and urban outdoor environments.” Seite 1
- Literaturrecherche
- Erwartete Ergebnisse
- GF4Shape method
Textstelle · Seite 1
- “Develop a method, GF4Shape, that fuses geometric features and learned features based on the diffusion approach.” Seite 1
- basic implementation
Textstelle · Seite 2
- “Achieve a basic implementation” Seite 2
- geometric-feature-based regularizer
Textstelle · Seite 1
- “Design geometric-feature-based regularizer for shape completion and compare to the M1 results.” Seite 1
- report
Textstelle · Seite 2
- “Report writing” Seite 2
- final presentation
Textstelle · Seite 2
- “Final presentation” Seite 2
- GF4Shape method
Voraussetzungen
- Erforderliche Kenntnisse
- Im PDF nicht angegeben
- Wünschenswerte Kenntnisse
- Im PDF nicht angegeben
- Geeignete Studienrichtungen
- Im PDF nicht angegeben
- Programmierung im Projekt
- Teilweise
Textstelle · Seite 2
- “Achieve a basic implementation” Seite 2
- Programmierkenntnisse als Voraussetzung
- Im PDF nicht angegeben
Rahmenbedingungen
- Teamgröße
- a selection of students from different TUM departments
Textstelle · Seite 2
- “The group of students for this project is a selection of students from different TUM departments” Seite 2
- Dauer
- 13.04.2026 - 31.08.2026
- Beginn
- 13.04.2026Normalisiertes Datum: 2026-04-13
Textstelle · Seite 1
- “13.04.2026 - 14.05.2026” Seite 1
- Bewerbungsfrist
- Im PDF nicht angegeben
- Arbeitsort
- Im PDF nicht angegeben
- Arbeitsmodus
- Im PDF nicht angegeben
- Arbeitssprache
- data analysis and mathematics
Textstelle · Seite 2
- “the common language to all students is data analysis and mathematics.” Seite 2
- data analysis and mathematics
Lernen und Betreuung
- Lernmöglichkeiten
- Students not directly related to data science will benefit by attending Fundamentals of Mathematical Statistics to approach interdisciplinary work
Textstelle · Seite 2
- “Students not directly related to data science will benefit by attending the course Fundamentals of Mathematical Statistics in order to approach interdisciplinary work.” Seite 2
- Students not directly related to data science will benefit by attending Fundamentals of Mathematical Statistics to approach interdisciplinary work
- Angebotene Betreuung
- Im PDF nicht angegeben
Bewerbung und Kontakt
- Kontaktpersonen
- Prof. Dr. Drton
Textstelle · Seite 1
- “Project leads: Prof. Dr. Drton (TUM Department of Mathematics)” Seite 1
- Dr. Ricardo Cabra
Textstelle · Seite 1
- “Dr. Ricardo Cabra (Munich Data Science Institute MDSI)” Seite 1
- Dr. Yan Xia
Textstelle · Seite 1
- “Scientific lead: Dr. Yan Xia and Prof. Dr. Olaf Wysocki” Seite 1
- Prof. Dr. Olaf Wysocki
Textstelle · Seite 1
- “Scientific lead: Dr. Yan Xia and Prof. Dr. Olaf Wysocki” Seite 1
- Prof. Dr. Drton
- Bewerbung
- Im PDF nicht angegeben
- Weitere Informationen
- Im PDF nicht angegeben