Computer Vision Decoded: Recent Episodes
EveryPoint
A tidal wave of computer vision innovation is quickly having an impact on everyone's lives, but not everyone has the time to sit down and read through a bunch of news articles and learn what it means for them. In Computer Vision Decoded, we sit down with Jared Heinly, the Chief Scientist at EveryPoint, to discuss topics in today’s quickly evolving world of computer vision and decode what they mean for you. If you want to be sure you understand everything happening in the world of computer vision, don't miss an episode!
In this episode of Computer Vision Decoded, we are going to dive into one of the hottest topics in the industry: Neural Radiance Fields (NeRFs)
We are joined by Matt Tancik, a student pursuing a PhD in the computer science and electrical engineering department at UC Berkeley. He has also contributed research to the original NeRF project in 2020 along with several others since then.
Last but not least, he is building NeRFStudio - a collaboration friendly studio for NeRFs.
In this episode you will learn about what NeRFs are and more importantly what they are not. Matt goes into the challenges of large scale NeRF creation with his experience with Block-NeRF.
Follow Matt's work at https://www.matthewtancik.com/
Get started with Nerfstudio here: https://docs.nerf.studio/en/latest/
Block-NeRF details: https://waymo.com/research/block-nerf/
00:00 Intro
00:45 Matt’s Background Into NeRF Research
04:00 What is a NeRF and how it is different from photogrammetry
11:57 Can geometry be extracted from NeRFs?
15:30 Will NeRFs supersede photogrammetry in the future?
22:47 Block-NeRF and the pros and cons of using 360 cameras
25:30 What is the goal of Block-NeRF
30:44 Why do NeRFs need large GPUs to compute?
35:45 Meshes to simulate NeRF visualizations
40:28 What is Nerfstudio?
47:40 How to get started with Nerfstudio
Follow Jared Heinly on Twitter: https://twitter.com/JaredHeinly
Follow Jonathan Stephens on Twitter at: https://twitter.com/jonstephens85
This episode is brought to you by EveryPoint. Learn more about how EveryPoint is building an infinitely scalable data collection and processing platform for the next generation of spatial computing applications and services: https://www.everypoint.io
In this episode of Computer Vision Decoded, we are going to dive into image capture best practices for 3D reconstruction.
At the end of this livestream, you will have learned the basics for capturing scenes and objects. We will also provide a downloadable visual guide for reference on your next 3D reconstruction project.
Download the official guide here to follow along: https://tinyurl.com/4n2wspkn
00:00 Intro
04:40 Camera motion overview
07:15 Good camera motions
18:43 Transition camera motions
30:39 Bad camera motions
39:27 How to combine camera motions
49:16 Loop Closure
57:42 Image Overlap
1:14:00 Lighting and camera gear
Watch out episode of Computer Vision in the Wild to learn more about capturing images outside and in busy locations: https://youtu.be/FwVBR6KFjPI
Follow Jared Heinly on Twitter: https://twitter.com/JaredHeinly
Follow Jonathan Stephens on Twitter at: https://twitter.com/jonstephens85
This episode is brought to you by EveryPoint. Learn more about how EveryPoint is building an infinitely scalable data collection and processing platform for the next generation of spatial computing applications and services: https://www.everypoint.io
In this episode of Computer Vision Decoded, we join Jared Heinly and Jonathan Stephens from EveryPoint for their live reaction to the iPhone 14 series announcement. They go in depth into what all the camera specs mean to the average person. We also explain basics of computational photography and how Apple is able to get great photos from a small camera sensor.
00:00 Intro
02:43 Apple Watch Review
06:58 Airpods Pro Review
09:40 iPhone 14 Initial Reaction
15:05 iPhone 14 Camera Specs Breakdown
37:13 iPhone 14 Pro Initial Reaction
40:47 iPhone 14 Pro Camera Specs Breakdown
Follow Jared Heinly on Twitter
Follow Jonathan Stephens on Twitter
This episode is brought to you by EveryPoint. Learn more about how EveryPoint is building an infinitely scalable data collection and processing platform for the next generation of spatial computing applications and services: https://www.everypoint.io
In this episode of Computer Vision Decoded, we sit down with Jared Heinly, Chief Scientist at EveryPoint, to discuss 3D reconstruction in the wild. What does “in the wild” mean? This means 3D reconstructing objects and scenes in non-controlled environments where you may have limitations with lighting, access, reflective surfaces, etc.
00:00 Intro
01:30: What are Duplicate Scene Structures and How to Avoid Them
14:30: How Jared used 100 million crowdsourced photos to 3d reconstruct 12,903 landmarks
27:10: The benefits of capturing video for 3D reconstruction
31:30: The benefits of using a drone to capture stills for 3D reconstruction
34:20: Considerations for using installed cameras for 3d reconstruction
38:30: How to work with sun issues
44:25: Determining how far from the object you should be when capturing images
50:35: How to capture objects with reflective surfaces
53:40: How work around scene obstructions
57:20: What cameras you should use
Jared Heinly’s Academic Papers and Projects
Paper: Correcting the Duplicate Scene Structure In Sparse 3D Reconstruction
Project: Reconstructing the World in Six Days
Video: Reconstructing the world in Six Days
Follow Jared Heinly on Twitter
Follow Jonathan Stephens on Twitter
This episode is brought to you by EveryPoint. Learn more about how EveryPoint is building an infinitely scalable data collection and processing platform for the next generation of spatial computing applications and services: https://www.everypoint.io
In this episode of Computer Vision Decoded we dive into Jared Heinly's recent trip to the CVPR Conference. We cover: what the conference about, who should attend, what are the emerging trends in computer vision, how machine learning is being used in 3D reconstruction, and what NeRFs are for.
00:00 - Introduction
00:36 - What is CVPR?
02:49 - Who should attend CVPR?
08:11 - What are emerging trends in Computer Vision?
14:34 - What is the value of NeRFs?
20:55 - How should you attend as a non-scientist or academic?
Follow Jared Heinly on TwitterFollow Jonathan Stephens on TwitterCVPR Conference
Episode sponsored by: EveryPoint
In this inaugural episode of Computer Vision Decoded we dive into the recent announcements at WWDC 2022 and find out what they mean for the computer vision community. We talk about what Apple is doing with their new RoomPlan API and how computer vision scientists can leverage it for better experiences. We also cover the enhancements to video and photo capture during an active ARKit Session.
00:00 - Introduction
00:25 - Meet Jared Heinly
02:10 - RoomPlan API
06:23 - Higher Resolution Video with ARKit
09:17 - The importance of pixel size and density
13:13 - Copy and Paste Objects from Photos
16:47 - CVPR Conference Overview
Follow Jared Heinly on TwitterFollow Jonathan Stephens on TwitterLearn about RoomPlan API OverviewLearn about ARKit 6 HighlightsCVPR Conference
Episode sponsored by: EveryPoint