Jon Barron - Understanding and Extending Neural Radiance Fields
October 13, 2020. MIT-CSAIL
Abstract: Neural Radiance Fields (Mildenhall, Srinivasan, Tancik, et al., ECCV 2020) are an effective and simple technique for synthesizing photorealistic novel views of complex scenes by optimizing an underlying continuous volumetric radiance field, parameterized by a (non-convolutional) neural network. I will discuss and review NeRF and then introduce two works that closely relate to it: First, I will explain why NeRF (and other CPPN-like architectures that map from low-dimensional coordinates to intensities) depend critically on the use of a trigonometric "positional encoding", aided by insights provided by the neural tangent kernel literature. Second, I will show how NeRF can be extended to incorporate explicit reasoning about occluders and appearance variation, and can thereby enable photorealistic view synthesis and photometric manipulation using only unstructured image collections.
Bio: Jon Barron is a staff research scientist at Google, where he works on computer vision and machine learning. He received a PhD in Computer Science from the University of California, Berkeley in 2013, where he was advised by Jitendra Malik, and he received a Honours BSc in Computer Science from the University of Toronto in 2007. He received a National Science Foundation Graduate Research Fellowship in 2009, the C.V. Ramamoorthy Distinguished Research Award in 2013, the PAMI Young Researcher Award in 2020, and the ECCV Best Paper Honorable Mention in both 2016 and 2020.
Видео Jon Barron - Understanding and Extending Neural Radiance Fields канала Vision & Graphics Seminar at MIT
Abstract: Neural Radiance Fields (Mildenhall, Srinivasan, Tancik, et al., ECCV 2020) are an effective and simple technique for synthesizing photorealistic novel views of complex scenes by optimizing an underlying continuous volumetric radiance field, parameterized by a (non-convolutional) neural network. I will discuss and review NeRF and then introduce two works that closely relate to it: First, I will explain why NeRF (and other CPPN-like architectures that map from low-dimensional coordinates to intensities) depend critically on the use of a trigonometric "positional encoding", aided by insights provided by the neural tangent kernel literature. Second, I will show how NeRF can be extended to incorporate explicit reasoning about occluders and appearance variation, and can thereby enable photorealistic view synthesis and photometric manipulation using only unstructured image collections.
Bio: Jon Barron is a staff research scientist at Google, where he works on computer vision and machine learning. He received a PhD in Computer Science from the University of California, Berkeley in 2013, where he was advised by Jitendra Malik, and he received a Honours BSc in Computer Science from the University of Toronto in 2007. He received a National Science Foundation Graduate Research Fellowship in 2009, the C.V. Ramamoorthy Distinguished Research Award in 2013, the PAMI Young Researcher Award in 2020, and the ECCV Best Paper Honorable Mention in both 2016 and 2020.
Видео Jon Barron - Understanding and Extending Neural Radiance Fields канала Vision & Graphics Seminar at MIT
Показать
Комментарии отсутствуют
Информация о видео
26 февраля 2021 г. 21:57:22
00:54:44
Другие видео канала
BARF 🤮: Bundle-Adjusting Neural Radiance Fields (ICCV 2021 oral)Matthias Niessner - Why Neural Rendering is Super Cool!Deformable Neural Radiance FieldsNeRF: Representing Scenes as Neural Radiance Fields for View Synthesis (ML Research Paper Explained)CSC2547 DeepSDF Learning Continuous Signed Distance Functions for Shape RepresentationNVIDIA’s New AI: Wow, Instant Neural Graphics! 🤖NeRF in the Wild: Neural Radiance Fields for Unconstrained Photo Collections [updated]Block-NeRFFourier Feature Networks and Neural Volume RenderingTUM AI Lecture Series - Understanding and Extending Neural Radiance Fields (Jonathan T. Barron)Mip-NeRF: A Multiscale Representation for Anti-Aliasing Neural Radiance Fields[ECCV 2020] NeRF: Neural Radiance Fields (10 min talk)Total Relighting SIGGRAPH Talk (Full Length)Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance FieldsAdvances in Neural Rendering (SIGGRAPH 2021 Course) Part 1 of 2FastNeRF: High-Fidelity Neural Rendering at 200FPS [Condensed]Math's Fundamental FlawMatthew Tancik: Neural Radiance Fields for View SynthesisImplicit Neural Representations: From Objects to 3D ScenesVincent Sitzmann: Implicit Neural Scene Representations