CVPR 2020 Six-hour Tutorial Is Online! Cutting-edge Methods for Generating New Perspectives
As one of the world's top three conferences in the field of computer vision, CVPR attracts many professionals every year. CVPR 2020 has attracted more than 10,000 submissions. Recently, the CVPR 2020 conference was held online, and nearly 30 tutorials have been released. Let's attend the conference online.
CVPR 2020 was originally scheduled to be held at the Seattle Convention Center in Washington State from June 16 to 20, but due to the severe global epidemic situation, CVPR 2020 officials announced at the beginning of the year that the conference will be held entirely online.
Before the main event, workshops and teaching courses will be held from June 14th to 16th, which will be conducted independently online by the organizers of each section.
Nearly thirty teaching courses are already in progress. Some courses are currently broadcast live on YouTube, and some courses are in the form of ZOOM conferences. Some of them have been recorded and shared online.
Generating new perspectives: from the past to the future
Today's recommended tutorials Novel View Synthesis: From Depth-Based Warping to Multi-Plane Images and BeyondHosted by NVIDIA, live streamed via YouTube, and course content uploaded.
New perspective synthesisIt is a long-standing challenge at the intersection of computer graphics and computer vision.
The earliest research in this field can be traced back to the 1990s. Early methods for synthesizing new perspectives were to interpolate between corresponding pixels of the input image or between rays in space.
In recent years, deep learning methods have greatly improved the quality of research and brought new scientific research directions to this field.
Course Arrangement & Organizers
In this tutorial, we first introduced the principles of new perspective generation, including providing background and classification of different methods, and invited research teams with the latest methods in this field to share their experiences.
| course | Video Location | lecturer |
| A review of new perspective synthesis | 00:00-29:35 | Orazio GalloData Scientist, NVIDIA |
| Thinking based on image rendering | 31:30-1:08:25 | Rick Szeliski Director of Facebook Image Computing Team |
| SynSin: Single Image View Synthesis | 1:08:45-1:31:57 | Olivia Wiles PhD, University of Oxford |
| Generating new perspectives with multi-plane images | 1:42:32-2:15:30 | Richard Tucker |
| Generate new perspectives through VR devices | 2:17:00-2:45:30 | Gaurav Chaurasia Oculus |
| View and frame interpolation for light field cameras | 3:30:50-3:58:30 | Nima Kalantari Texas A&M |
| NeRF: New Perspective Synthesis of Scenes Represented with Neural Radiance Fields | 3:59:40-4:25:40 | Pratul Srinivasan UC Berkeley |
| New perspective synthesis in dynamic scenes | 4:30:15-4:53:45 | Jae Shin Yoon Minnesota State University |
| Roundtable Discussion | 5:51:35 - End | All lecturers |
Easter egg: Slang comparison table
Perspective conversion perspective transformation
Affine transformation Affine transformation
Homography homography
Inverse homography inverse homography
Depth Image-Based Rendering DIBR (depth image-based rendering)
View generation from a single image SIVG (Single-image-based view generation)
Thinning the network refinement network
Backward optical flow map backward flow map
-- over--