CS Other Presentations

Department of Computer Science - University of Cyprus

Besides Colloquiums, the Department of Computer Science at the University of Cyprus also holds Other Presentations (Research Seminars, PhD Defenses, Short Term Courses, Demonstrations, etc.). These presentations are given by scientists who aim to present preliminary results of their research work and/or other technical material. Other Presentations serve as a forum for educating Computer Science students and related announcements are disseminated to the Department of Computer Science (i.e., the csall list):

RSS Directions

Presentations Coordinator: Demetris Zeinalipour

Presentation: Graph-based Neural Network Models for 3D Shape Segmentation, Mr. Marios Loizou (University of Cyprus, Cyprus), Friday, April 26, 2024, 14:30-15:30 EET.

The Department of Computer Science at the University of Cyprus cordially invites you to the Presentation entitled:

Graph-based Neural Network Models for 3D Shape Segmentation

Abstract:
The segmentation of 3D shapes into their constituent parts has been a long-standing problem in Computer Vision and Graphics. Recent breakthroughs in 3D Deep Learning led to numerous methods for learning effective representations useful for high-level shape processing tasks, including shape segmentation. Despite these significant advances, most methods rely on processing local geometric neighborhoods and often disregard broader context, such as structure, symmetries, and correspondences with other shapes that are often useful for discovering and extracting parts in geometric shape representations. Moreover, commonly used 3D datasets comprise mainly man-made objects with simple structure and lack large-scale models with high structural complexity. This thesis presents graph-based neural methods that model complex structural and spatial relations within the same shape as well as across shapes in their graph representation to produce more consistent and accurate shape segmentations. Additionally, the thesis introduces the first publicly available large-scale dataset of annotated 3D building models. Buildings have more challenging structural complexity compared to objects in common benchmarks, thus, the dataset serves as a useful benchmark to evaluate segmentation algorithms on large-scale, structurally complex geometric data. The thesis concludes with a discussion for future research directions in shape segmentation, such as leveraging self-supervised pre-training procedures, open-vocabulary models, and unsupervised structure learning for further improving graph-based approaches for segmentation as well as 3D shape and scene understanding in general.

Short Bio:
Marios Loizou is a Ph.D. candidate in the Computer Science Department at the University of Cyprus (UCY), supervised by Prof. Evangelos Kalogerakis (UMass Amherst) and Prof. Yiorgos Chrysanthou. Concurrently, he serves as a Research Associate at the CYENS Centre of Excellence under the guidance of Dr. Melinos Averkiou. His research lies at the intersection of deep learning, computer vision, and geometry processing, focusing on 3D shape segmentation and analysis. He is particularly interested in developing geometric deep learning architectures for processing 3D shape representations and inferring real-world semantics. Prior to his current pursuits, he obtained a Dipl.-Ing. in Computer Engineering and Informatics from the University of Patras, Greece, and an MSc in Computer Science from the University of Cyprus.

	Other Presentations Web: https://www.cs.ucy.ac.cy/colloquium/presentations.php
	Colloquia Web: https://www.cs.ucy.ac.cy/colloquium/
	Calendar: https://www.cs.ucy.ac.cy/colloquium/schedule/cs.ucy.pres.2024.loizou.ics