Google Maps Getting More Immersive and Stable

As mentioned in todays event ... 

New ways Maps is getting more immersive and sustainable

Feb 08, 2020 Chris Phillips, VP & General Manager, Geo

Last year, we shared our vision for the future of Google Maps — an immersive, intuitive map that reimagines how you explore and navigate, while helping you make more sustainable choices. Today we’re demonstrating how AI is bringing this vision to life, with updates for immersive view and Live View, along with new features for electric vehicle (EV) drivers and people who walk, bike or ride public transit.

Immersive view: rolling out now 🎉

Immersive view is an entirely new way to explore a place — letting you feel like you’re right there, even before you visit. Using advances in AI and computer vision, immersive view fuses billions of Street View and aerial images to create a rich, digital model of the world. And it layers helpful information on top like the weather, traffic, and how busy a place is.

Say you’re planning a visit to the Rijksmuseum in Amsterdam. You can virtually soar over the building and see where things like the entrances are. With the time slider, you can see what the area looks like at different times of day and what the weather will be like. You can also spot where it tends to be most crowded so you can have all the information you need to decide where and when to go. If you’re hungry, glide down to the street level to explore nearby restaurants — and even take a look inside to quickly understand the vibe of a spot before you book your reservation.

To create these true-to-life scenes, we use neural radiance fields (NeRF), an advanced AI technique, transforms ordinary pictures into 3D representations. With NeRF, we can accurately recreate the full context of a place including its lighting, the texture of materials and what’s in the background. All of this allows you to see if a bar’s moody lighting is the right vibe for a date night or if the views at a cafe make it the ideal spot for lunch with friends.

Immersive view starts rolling out today in London, Los Angeles, New York, San Francisco and Tokyo. And in the coming months, it’ll launch in even more cities, including in Amsterdam, Dublin, Florence and Venice. ... ' 

New X-Ray Technique Could Help Detect Explosives, Tumors

 Shape detection leveraged.  

New X-Ray Technique Could Help Detect Explosives, Tumors

By Adrianna Nine on September 15, 2022 at 1:02 pm 

A new X-ray technique that combines conventional equipment with a deep-learning algorithm might find its way into both security settings and the healthcare industry.

Researchers from the United Kingdom’s University College London (UCL) recognized that X-ray security systems, though good at detecting shapes, weren’t so great at recognizing textures. Identifying textural abnormalities could be the key to locating explosives and other harmful items—especially those hidden away within larger objects. So they set about devising a system that could be paired with existing equipment to detect concerning textures. ... ' 

Seeing the way we do

New ways of perceptive seeing, now with Texture and Shape

GLOM: Teaching Computers to See the Way(s) We Do  By John Delaney,  Commissioned by CACM Staff   September 14, 2021

At the virtual Collision technology  earlier this year deep learning pioneer Geoffrey Hinton explained how he conceived of a new type of neural network that, he said, would be able to perceive things the way people do.

Hinton, an emeritus distinguished professor in the department of computer science of the Faculty of Arts & Science at Canada's University of Toronto, and also an Engineering Fellow at Google, is responsible for some of the biggest breakthroughs in deep learning and neural networks. He was honored as co-recipient of the 2018 ACM A.M. Turing Award, along with Yoshua Bengio and Yann LeCun, for conceptual and engineering breakthroughs that have made deep neural networks a critical component of modern computing.

In Hinton's Collision talk, he pointed out that the representations used by most neural networks performing object classification are produced by convolutional neural networks, which work well at classifying objects such as images or words, even winning competitions such as the ImageNet Large Scale Visual Recognition Challenge, but they perceive images in a very different way than people do, which can sometimes lead to "crazy errors."

"They use lots of texture information, which people are insensitive to," Hinton said, "but they fail to use a lot of shape information, which people are very sensitive to."  .... '

Automating Material-Matching for Movies and Video Games

Towards an automation of many kinds of image context.

Automating Material-Matching for Movies and Video Games

MIT News by Adam Conner-Simons

Massachusetts Institute of Technology (MIT) and Adobe researchers described a new system they developed for texturing computer-generated objects as being easy as capturing a photo of an object and reconstructing it on a laptop. MaTCH uses a "DiffMat" library that supplies the various building blocks for constructing different textured materials, while dozens of "procedural graphs" comprise different nodes that convert input into output in specific artistic ways. Said Shi, "The neural network selects the most appropriate combinations of filter nodes until it perceptually matches the appearance of the user's input image." The team tested MaTCH on rendered synthetic materials and real materials captured on camera, and found the system can recreate materials more accurately and at higher resolution than current state-of-the-art methods.  ... "