• Radu Timofte, University of Wurzburg and ETH Zurich,
• Andrey Ignatov, AI Benchmark and ETH Zurich,
• Ren Yang, ETH Zurich,
• Marcos V. Conde, University of Wurzburg,
• Furkan Kınlı, Özyeğin University,

• Codruta Ancuti, UPT
• Boaz Arad, Ben-Gurion University of the Negev
• Siavash Arjomand Bigdeli, DTU
• Michael S. Brown, York University
• Jianrui Cai, The Hong Kong Polytechnic University
• Chia-Ming Cheng, MediaTek
• Cheng-Ming Chiang, MediaTek
• Sunghyun Cho, Samsung
• Marcos V. Conde, University of Wurzburg
• Chao Dong, SIAT
• Weisheng Dong, Xidian University
• Touradj Ebrahimi, EPFL
• Paolo Favaro, University of Bern
• Graham Finlayson, University of East Anglia
• Corneliu Florea, University Politechnica of Bucharest
• Bastian Goldluecke, University of Konstanz
• Shuhang Gu, OPPO & University of Sydney
• Christine Guillemot, INRIA
• Felix Heide, Princeton University & Algolux
• Chiu Man Ho, OPPO,
• Hiroto Honda, Mobility Technologies Co Ltd.
• Andrey Ignatov, ETH Zurich
• Aggelos Katsaggelos, Northwestern University
• Jan Kautz, NVIDIA
• Furkan Kınlı, Özyeğin University
• Christian Ledig, University of Bamberg
• Seungyong Lee, POSTECH
• Kyoung Mu Lee, Seoul National University
• Juncheng Li, The Chinese University of Hong Kong
• Yawei Li, ETH Zurich
• Stephen Lin, Microsoft Research
• Guo Lu, Beijing Institute of Technology
• Kede Ma, City University of Hong Kong
• Vasile Manta, Technical University of Iasi
• Rafal Mantiuk, University of Cambridge
• Zibo Meng, OPPO
• Yusuke Monno, Tokyo Institute of Technology
• Hajime Nagahara, Osaka University
• Vinay P. Namboodiri, University of Bath/li>
• Federico Perazzi, Bending Spoons
• Fatih Porikli, Qualcomm CR&D
• Antonio Robles-Kelly, Deakin University
• Aline Roumy, INRIA
• Christopher Schroers, Disney Research | Studios
• Nicu Sebe, University of Trento
• Eli Shechtman, Creative Intelligence Lab at Adobe Research
• Gregory Slabaugh, Queen Mary University of London
• Sabine Süsstrunk, EPFL
• Yu-Wing Tai, Kuaishou Technology & HKUST
• Robby T. Tan, Yale-NUS College
• Masayuki Tanaka, Tokyo Institute of Technology
• Hao Tang, ETH Zurich
• Qi Tian, Huawei Cloud & AI
• Radu Timofte, University of Wurzburg & ETH Zurich
• George Toderici, Google
• Luc Van Gool, ETH Zurich & KU Leuven
• Longguang Wang, National University of Defense Technology
• Yingqian Wang, National University of Defense Technology
• Gordon Wetzstein, Stanford University
• Ming-Hsuan Yang, University of California at Merced & Google
• Ren Yang, ETH Zurich
• Wenjun Zeng, Microsoft Research
• Kai Zhang, ETH Zurich
• Yulun Zhang, ETH Zurich
• Jun-Yan Zhu, Carnegie Mellon University
• Wangmeng Zuo, Harbin Institute of Technology

Sabine Süsstrunk

EPFL

Title: Uncovering local semantics in CNNs and GANs

Abstract: Automatically localizing similar semantic concepts within an image or a set of images allows for many applications, such as image segmentation, localization, and image editing. We propose Deep Feature Factorization (DFF), a method capable of detecting hierarchical cluster structures in feature space of a convolutional neural network (CNN). These clusters are visualized as heat maps, which highlight semantically matching regions across a set of images, revealing what the network `perceives' as similar. Analogue structures can also be found in generative adversarial networks (GANs). Focusing on StyleGAN, we introduce two simple and effective methods for making local, semantically-aware edits to a GAN output image. Our methods require neither supervision from an external model, nor involve complex spatial morphing operations. Semantic editing is demonstrated on variety of scenes, and even real photographs using GAN inversion.

Bio: Sabine Süsstrunk is Full Professor and Director of the Image and Visual Representation Lab in the School of Computer and Communication Sciences (IC) at the Ecole Polytechnique Fédérale (EPFL), Lausanne, Switzerland. Her main research areas are in computational imaging, computer vision, machine learning, and computational image quality and aesthetics. She is President of the Swiss Science Council SSC, Founding Member and Member of the Board of the EPFL-WISH (Women in Science and Humanities) Foundation, board member of the SRG SSR (Swiss Radio and Television Corporation), and Co-Founder and board member of Largo Films, Ltd. Sabine is a Fellow of IEEE, IS&T, and the Swiss Academy of Engineering Sciences (SATW).

Felix Heide

Princeton University & Algolux

Title: The Differentiable Camera: Designing Cameras to Detect the Invisible

Abstract: Although today's cameras fuel diverse applications, from personal photography to self-driving vehicles, they are designed in a compartmentalized fashion where the optics, sensor, image processing pipeline, and vision models are often devised in isolation: a camera design is decided by intermediate metrics describing optical performance, signal to noise ratio, and image quality, even though object detection scores may only matter for the camera application. In this talk, I will present a differentiable camera architecture, including compound optics, sensing and exposure control, image processing, and downstream vision models. This architecture allows us to learn cameras akin to neural networks, entirely guided by downstream loss functions. Learned cameras move computation to the optics, with entirely different optical stacks for different vision tasks (and beating existing stacks such as Tesla's Autopilot). The approach allows us to learn entirely new domain-specific cameras that perform imaging and vision tasks jointly, and learn active illumination together with the image pipeline, achieving accurate dense depth and vision tasks in heavy fog, snow, and rain (beating scanning lidar methods). Finally, I will describe an approach that makes the scene itself differentiable, allowing us to backpropagate gradients through the entire capture and processing chain in an inverse rendering fashion. As such, the proposed novel breed of learned cameras brings unprecedented capabilities in optical design, imaging, and vision.

Bio: Felix Heide is an Assistant Professor at Princeton University and Co-Founder and Chief Technology Officer of self-driving vehicle startup Algolux. He is researching the theory and application of computational imaging. As such, Felix's work lies at the intersection of optics, machine learning, optimization, computer graphics, and computer vision. Felix received his Ph.D. from the University of British Columbia. He obtained his MSc from the University of Siegen and was a postdoc at Stanford University. His doctoral dissertation won the Alain Fournier Ph.D. Dissertation Award and the SIGGRAPH outstanding doctoral dissertation award. He won an NSF CAREER Award and Sony Young Faculty Award 2021. He co-founded the autonomous driving startup Algolux, and his research is directly used by Waymo, Cruise, and Google as part of their autonomous vehicle programs.

Gordon Wetzstein

Stanford University

Title: Efficient Neural Scene Representation, Rendering, and Generation

Abstract: Neural radiance fields and scene representation networks offer unprecedented capabilities for photorealistic scene representation, view interpolation, and many other tasks. In this talk, we discuss expressive scene representation network architecture, efficient neural rendering approaches, and generalization strategies that allow us to generate photorealistic multi-view-consistent humans or cats using state-of-the-art 3D GANs.

Bio: Gordon Wetzstein is an Associate Professor of Electrical Engineering and, by courtesy, of Computer Science at Stanford University. He is the leader of the Stanford Computational Imaging Lab and a faculty co-director of the Stanford Center for Image Systems Engineering. At the intersection of computer graphics and vision, artificial intelligence, computational optics, and applied vision science, Prof. Wetzstein's research has a wide range of applications in next-generation imaging, wearable computing, and neural rendering systems. Prof. Wetzstein is the recipient of numerous awards, including an NSF CAREER Award, an Alfred P. Sloan Fellowship, an ACM SIGGRAPH Significant New Researcher Award, a Presidential Early Career Award for Scientists and Engineers (PECASE), an SPIE Early Career Achievement Award, an Electronic Imaging Scientist of the Year Award, an Alain Fournier Ph.D. Dissertation Award as well as many Best Paper and Demo Awards.

Bjorn Ommer

University of Munich

Title: Stable Diffusion++: Democratizing Visual Synthesis

Abstract: The ultimate goal of computer vision are models that can understand our (visual) world. Recent deep generative models for visual synthesis open up new avenues towards scene understanding by providing accurate representations of both, the rich details and the diversity featured by large, heterogeneous image datasets. Still, they exhibit specific limitations that restrict their applicability and performance especially in complex tasks such as high-resolution image synthesis. We will discuss a solution, latent diffusion models a.k.a. "Stable Diffusion", that significantly improves the efficiency of diffusion models. Now billions of training samples can be summarized in compact representations of just a few gigabyte so that the approach runs on even consumer GPUs, thus making high-quality visual synthesis accessible for everyone. We will then discuss ongoing work that shows what we should better NOT learn to represent with generative models and we instead propose a semi-parametric approach.

Bio: Björn Ommer is a full professor at the University of Munich where he is heading the Machine Vision and Learning Group. Before he was a full professor in the department of mathematics and computer science at Heidelberg University. He received his diploma in computer science from University of Bonn and his PhD from ETH Zurich. Thereafter, he was a postdoc in the vision group of Jitendra Malik at UC Berkeley. Björn serves as an associate editor for IEEE T-PAMI. His research interests include semantic scene understanding, visual synthesis and retrieval, self-supervised metric and representation learning, and explainable AI. Moreover, he is applying this basic research in interdisciplinary projects within the digital humanities and the life sciences.

Bo Zhu

Meta Reality Labs

Title: Practical considerations for on-device denoising

Abstract: This talk addresses some practical considerations of ML-based denoising for on-device contexts. We will discuss rapid and inexpensive methods for photon transfer curve characterization without the need for flat field illuminators and integration spheres, the importance of variance stabilization transforms in quantized, computationally efficient models, as well as compression of raw bayer data for offloaded inference scenarios.

Bio: Bo Zhu is a Research Scientist Manager of AI at Meta Reality Labs, leading a computer vision research team to advance imaging performance in AR & VR devices. Prior to joining Meta, Bo was cofounder and CTO of BlinkAI, a spinoff from ML image reconstruction research (Nature, 2018) he led as a postdoc at Harvard, whose AI video denoising products were commercially deployed on tens of millions of devices globally (including DXOMark’s #1 ranked camera phone in 2021, Mi 11 Ultra) and awarded 2021 Product of the Year at Embedded Vision Summit. He obtained his S.B., M.Eng., and Ph.D. from MIT and performed research at the Martinos Center for Biomedical Imaging and CSAIL.

Pratul Srinivasan

Google Research

Title: Neural Inverse Rendering

Abstract: Recent progress in neural 3D object and scene representations such as Neural Radiance Fields (NeRFs) are bringing us closer to tackling the longstanding inverse rendering problem of jointly estimating geometry, materials, and lighting from observed images. In this talk, I will discuss how current neural field approaches to inverse rendering can be thought of as lying on a spectrum of simulation vs. precomputation, and how this suggests potentially interesting and fruitful research directions.

Bio: Pratul Srinivasan is a research scientist at Google Research, where he works on problems at the intersection of computer vision and graphics. His recent work has focused on view synthesis, inverse rendering, and 3D reconstruction. He completed his PhD at UC Berkeley in 2020, where he was advised by Ravi Ramamoorthi and Ren Ng and supported by an NSF Graduate Fellowship. His research has been recognized by the David J. Sakrison Memorial Prize from UC Berkeley in 2020, the ACM Doctoral Dissertation Award Honorable Mention in 2022, the ECCV Best Paper Honorable Mention Award in 2020, the ICCV Best Paper Honorable Mention Award in 2021, and the CVPR Best Student Paper Honorable Mention Award in 2022.