June 11	June 12	June 13	June 14
Morning	Andrea Fusiello	Tanya Birch	Hannah Kerner	Zan Gojcic
Afternoon	Dalton Lunga	-	Konstantinos Karantzalos	-

Title: Graph Synchronization and Rigidity: Unraveling the Theory Underneath Structure from Motion

Abstract: In this talk, I will explore synchronization problems within graph theory, focusing on motion synchronization and its connection to Structure from Motion (SFM). The goal is to establish consistent orientations for a set of sensors based on relative motion estimates, addressing the challenges of noisy edge measurements. I will also touch on related localization issues, where unknown sensor positions and measurements such as distances or directions contribute to the complexity of the problem. The talk serves as a concise guide to understanding these intricacies, emphasizing the interplay between synchronization, SFM, and rigidity in computer vision and photogrammetry.

Bio: Andrea Fusiello earned his Laurea (M.S.) degree in computer science from the University of Udine (Italy) in 1994 and later completed his PhD in computer engineering from the University of Trieste in 1999. Following this, he served as a Research Fellow at Heriot-Watt University, Edinburgh, in 1999. From 2001 to 2011, he contributed significantly to the Department of Computer Science at the University of Verona. As an Associate Professor, he joined the DPIA at the University of Udine in 2012, ultimately achieving the position of Full Professor in 2023. With over 150 published articles and holding two patents, Andrea is a distinguished researcher. His broad research interests span various domains in Computer Vision, Photogrammetry, and Image Analysis, with a specific focus on 3-D modeling and reconstruction. He was honored with the Marr Prize - Honorable Mention in 2021.

Title: Frontiers in AI for Human Security

Abstract: The world is facing a growing crisis as natural disasters become more frequent and devastating, resulting in staggering loss of life and economic upheaval. In 2022 alone, natural catastrophes claimed over 30,000 lives and inflicted a financial toll of nearly $224 billion. With over 100+ TB of Earth data being collected daily, advances in AI are demonstrating tremendous progress in transforming data into life-saving actions.​ In this talk, we will share progress made by Oak Ridge National Laboratory in contributing to crisis management by creating AI that understands diverse Earth and climate data. Energy-efficient AI is becoming critical - we will show its applicability for instant extreme condition detection. We will further discuss how we are developing reliable multi-modal GeoAI foundation models to enable easy human-machine collaboration and support mission critical decisions, save lives and secure our future.

Bio: Dalton is currently a senior research scientist in machine learning-driven geospatial image analytics and a group leader for the geospatial-based artificial intelligence (GeoAI) group at ORNL. In this role, he leverages high-performance computing, machine learning, and computer vision to create foundational geospatial analytic methods enabling at-scale data generation that shed light on pattern-of-life and aid in crisis management. His efforts to generate accurate population estimates and information about urban growth and decline, for example, inform disaster response, identify at-risk areas, and address infrastructure mapping and monitoring. A Purdue University Ph.D. graduate and former employee at the council for scientific and industrial research in South Africa, Dalton brings backgrounds from academia, industry, program leadership, and extensive community service to ORNL to help advance geospatial-based scientific knowledge discovery for societal impact.

Title: Pixels with Purpose: Illuminating the Path to Change

Bio: Tanya Birch is a Senior Program Manager at Google, using Google's mapping technology, AI capabilities and Cloud platform for helping monitor ecosystems and biodiversity. She leads the Forest & Nature program within Geo Sustainability, which catalyzes positive environmental impact at scale using Google’s understanding of the real world. She works with leading public & private sector organizations in applying technology to address nature-based solutions to climate change. She led the program management of Dynamic World, a novel deep learning approach to land cover mapping with World Resources Institute, and is founding technology partner of Wildlife Insights, a global platform for biodiversity monitoring with 7 leading conservation organizations. Prior to Google, she researched and mapped human elephant conflict with the Sri Lanka Wildlife Conservation Society. She holds a BA in Geography and Environmental Studies from the University of California at Santa Barbara.

Title: Unlocking the Potential of Planetary-Scale Machine Learning for a Sustainable Future

Abstract: Remote sensing satellites capture peta-scale, multi-modal data capturing our dynamic planet across space, time, and spectrum. This rich data source holds immense potential for addressing local and planetary-scale challenges including food insecurity, poverty, climate change, and ecosystem preservation. Fully realizing this potential will require a new paradigm of machine learning approaches capable of tackling the unique character of remote sensing data. Machine learning approaches must be flexible enough to make use of the multi-modal multi-fidelity satellite data, process meter-scale observations over planetary scales, and generalize to the challenging diversity of remote sensing tasks. In this talk, I will present examples of how we are developing machine learning approaches for planetary data processing including self-supervised transformers for remote sensing data. I will also demonstrate how treating ML research and deployment as a unified approach instead of siloed steps leads to research advances that result in immediate societal impact, highlighting examples of how we are partnering directly with stakeholders to deploy our innovations in areas of critical need across the globe.

Bio: Hannah Kerner is an Assistant Professor of computer science in the School of Computing and Augmented Intelligence at Arizona State University. She is pioneering new machine learning techniques to harness the potential of remote sensing data to address global challenges like food insecurity and climate change. Her research aims to tackle barriers to realizing the benefits of machine learning in real-world applications that benefit society. As the AI Lead for NASA's agriculture programs, NASA Harvest and NASA Acres, she is deploying research methods in real applications across the globe; her projects have directly resulted in optimized agricultural planning, disaster response, and financial relief in various regions around the world. The impact of Kerner’s research was recognized in Forbes 30 Under 30 and the International Research Centre On Artificial Intelligence's Top 10 projects solving problems related to the UN's Sustainable Development Goals with AI.

Abstract: Nowadays, the marine environment concentrates significant amount of research and development aiming to map and monitor the sea surface, the water column, and the seabed with the same quality and efficiency as in land. Multimodal data, with increased spatial, spectral, temporal resolution acquired from different payloads onboard satellites, aerial, marine and submarine robots are delivering valuable information that should be exploited in an automated manner. In this talk, we will discuss cutting-edge sensing technologies, AI, ML, and data cloud/edge analytics that can harmonize observations, understand them, and monitor effectively the dynamic marine environment.

Bio: Dr. Karantzalos is a Professor at the National Technical University of Athens, joining the Remote Sensing Laboratory. His teaching and research interests include earth observation and remote sensing, geospatial big data and analytics, computer vision and machine learning. He has numerous publications in international journals and conferences and a number of awards and honors for his research contributions. He has more than 20 years of research experience, involved with more than 30 EU and national excellence/ competitive research projects. He is currently the Head of the Greek delegation in the European Space Agency and coordinates the design and implementation of the Greek small satellites program.

Abstract: Neural radiance fields have emerged as a powerful 3D representation, enabling photorealistic novel-view synthesis and reconstruction. However, the high visual fidelity comes at a high computational cost, as the volume rendering formulation requires many samples along each ray. On the other hand, surface-based rendering methods are very efficient to render but may lack high visual fidelity, especially in regions that are more fuzzy and volumetric. Recently, several works in the field of text-to-3D have shown how the two can be combined into a two-stage pipeline that yields improved visual fidelity and high-resolution details. In this talk, I will allude to further possibilities of combining the two formulations in order to get the best of both worlds.

Bio: Zan Gojcic is a senior research scientist and a research manager at the Nvidia Toronto Ai Lab, where his team is exploring topics related to neural reconstruction and data driven simulation. Prior to joining Nvidia, Zan has received his PhD from ETH Zurich. His thesis titled “Benefiting from local rigidity in 3D point cloud processing” was awarded ETH Medal - an award bestowed upon outstanding doctoral theses. During his PhD, Zan has also visited the geometric computation group led by Leonidas Guibas at Stanford University. His research interests are focused on 3D neural reconstruction, generative models, and their combination.