Four Invited Talks to GSP-CV 2021
Title: Graph signal processing: Theory and Algorithms (Prof. Antonio Ortega, USC Viterbi, USA)
Title: Graph Neural Networks and Their Applications to Point Cloud Processing (Prof. Enrico Magli, Politecnico di Torino, Italy)
This talk addresses the topic of graph neural networks in the framework of computer vision. Graph neural networks have witnessed increasing interest due to their ability to learn models of complex dependencies in a variety of fields. A brief overview of graph neural networks will be provided, along with examples of applications. Then the talk will focus on the topic of point cloud processing; this type of data is very important in many vision applications, and is well suited to graph neural networks. The talk will cover several applications including generative adversarial networks based on graph-convolutional layers, point cloud denoising networks, and point cloud inpainting methods. Finally a discussion of open research issues will be presented.
Biography: Enrico Magli is a Full Professor at Politecnico di Torino, where he leads a research group active in the field of deep learning for image and video processing. He is an Associate Editor of the IEEE Transactions on Circuits and Systems for Video Technology and the EURASIP Journal on Image and Video Processing, and a former Associate Editor of the IEEE Transactions on Multimedia. He is a Fellow of the IEEE and has been an IEEE Distinguished Lecturer. He is a co-recipient of the 2019 IEEE Multimedia Best paper award and the IEEE Geoscience and Remote Sensing Society 2011 Transactions Prize Paper Award. He has been general chair of IEEE ICME 2015 and IEEE MMSP 2013, and TPC chair / area chair for several conferences. He is co-founder and President of the ToothPic start-up, which markets cybersecurity solutions based on camera sensor fingerprints.
Title: Graph Signal Processing for Machine Learning: A Review and New Perspectives on Imaging (Dr. Dorina Thanou, EPFL, Switzerland)
The effective representation, processing, analysis, and visualization of large-scale structured data, especially those related to complex domains such as networks and graphs, are one of the key questions in modern machine learning. Graph signal processing (GSP), a vibrant branch of signal processing models and algorithms that aims at handling data supported on graphs, opens new paths of research to address this challenge. In this talk, we will highlight a few important contributions made by GSP concepts and tools, such as graph filters and transforms, to the development of novel machine learning algorithms, and their application in imaging. In particular, our discussion will focus on the following three aspects: exploiting data structure and relational priors, improving data and computational efficiency, and enhancing model interpretability. Finally, we will show some illustrative applications in imaging and computer vision.
Biography: Dorina Thanou is a senior scientist and lecturer at EPFL, leading the development of the Intelligent Systems for Medicine and Health research pillar, under the Center for Intelligent Systems. Prior to that, she was a Senior Data Scientist at the Swiss Data Science Centre. She got her M.Sc. and Ph.D. in Communication Systems and Electrical Engineering respectively, both from EPFL, Switzerland, and her Diploma in Electrical and Computer Engineering from the University of Patras, Greece. She is the recipient of the Best Student Paper Award at ICASSP 2015, and the co-author of the Best Paper Award at PCS 2016. Her research interests lie in the broader area of signal processing and machine learning and she is particularly interested in applying her expertise to intelligent systems for healthcare.
Title: Graph Moving Object Segmentation (Ass. Professor Sajid Javed, Khalifa University, UAE)
Moving Object Segmentation (MOS) is a fundamental task in computer vision. Due to undesirable variations in the background scene, MOS becomes very challenging for static and moving camera sequences. Several deep learning methods have been proposed for MOS with impressive performance. However, these methods show performance degradation in the presence of unseen videos; and usually, deep learning models require large amounts of data to avoid overfitting. In this talk, I will present the graph learning problem as a moving object segmentation problem. Graph learning has attracted significant attention in many computer vision applications since they provide tools to exploit the geometrical structure of data. In this talk, the concepts of graph signal processing are introduced for MOS. First, I will present a new algorithm that is composed of segmentation, background initialization, graph construction, unseen sampling, and a semi-supervised learning method inspired by the theory of recovery of graph signals. Secondly, theoretical developments are introduced, showing one bound for the sample complexity in semi-supervised learning, and two bounds for the condition number of the Sobolev norm. The algorithm has the advantage of requiring less labeled data than deep learning methods while having competitive results on both static and moving camera videos. The graph learning algorithm is also adapted for the video object segmentation tasks and is evaluated on six publicly available datasets outperforming several state-of-the-art methods in challenging conditions. Finally, I will draw a boundary to highlight its potential to solve other computer vision applications.
Biography: Sajid Javed is an Assistant Professor of Computer Vision in Electrical and Computer Engineering (ECE) department at Khalifa University of Science and Technology, UAE. Prior to that, he was a research scientist at Khalifa University Center for Autonomous Robotics System, UAE, from February 2019-April 2021. Before joining Khalifa University, he was a research fellow at the University of Warwick from October 2017 to December 2018. He received his B.Sc. (Hons) degree in computer science from the University of Hertfordshire, United Kingdom in 2010. After that, he completed his combined Master’s and Ph.D. studies in computer science from Kyungpook National University, South Korea, in August 2017. He is interested in computer vision, image processing, machine learning, and deep learning research problems. More specifically, he is working on background-foreground modeling, multiple object tracking, and single object tracking in video sequences. His research theme involves deep learning, robust principal component analysis, low-rank matrix completion, subspace learning, and unsupervised machine learning problems.