Disney’s deepfake algorithm
Disney Research and ETH Zurich introduced an algorithm for automatic face-swapping in photos and videos, which has a high enough resolution for filmmaking. It's the first deepfake method with a megapixel resolution, meaning it's convincing enough to alter faces for TV and movies, according to the researchers.
More:
- The goal is to use the neural-network-based method to substitute actors' faces, such as in the de-aging process or when an actor has died.
- Disney’s model can create deepfake videos with a 1024 x 1024 resolution, much higher than open-source model DeepFakeLab's 256 x 256 pixels.
- It's less time-consuming and expensive than Disney's current method for face-swapping, traditional VFX, which was used to re-create Carrie Fisher in Rogue One.
- TechCrunch's Darrell Etherington described the results as "a lot less nightmare fuel" than other deepfake attempts. He still noted that the subjects in the sample video were primarily white.
- The paper, “High-Resolution Neural Face Swapping for Visual Effects,” will be presented at this week's virtual 2020 Eurographics Symposium on Rendering.
See the full story here: http://studios.disneyresearch.com/2020/06/29/high-resolution-neural-face-swapping-for-visual-effects/
This German town replicated itself in VR to keep its tourism alive
Tourists may soon be able to explore the picturesque cross-timbered houses and historic churches of Herrenberg via virtual reality (VR), thanks to a digital twin developed with the High-Performance Computing Center Stuttgart (HLRS).
Building a digital twin
HLRS developed Herrenberg’s digital twin together with the Fraunhofer Institute, the University of Stuttgart, and Kommunikationsbüro Ulmer, starting with a concept called ‘space syntax’.
Dr. Fabian Dembski of HLRS said: “Just as the human skeleton provides a scaffolding for all of the other systems and functions of the human body, space syntax produces a 2D outline of physical grids in a city, offering a framework for performing spatial analysis, such as predicting the likely paths that car or pedestrian traffic might take to move from one point to another.”
We shouldn’t overestimate the technology, though, he noted. “Cities are not machines,” he commented. “A digital twin can be a great help in reducing complexity for cities, assessing measures at an early stage, and explaining interrelationships. But there are many aspects that are deeply human and cannot be reproduced in digital copies, such as culture, interpersonal relationships, joy, and happiness. A digital twin is a tool, not a solution.”
He added: “I also think it is important for cities to retain control over data and models,” urging co-operation between science, city administrations, businesses, and citizens.
See the full story here: https://thenextweb.com/syndication/2020/05/31/this-german-town-replicated-itself-in-vr-to-keep-its-tourism-alive/
Tom Hall interview — Why Doom co-creator is making virtual reality games
Tom Hall is a versatile game designer with over three decades of experience. He has worked on note games such as Doom, Duke Nukem, Rise of the Triad, Wolfenstein 3-D, and Anachronox (all traditional games). But he has also shown interest in the broader craft of making games for other platforms, working on the Diner Dash mobile games at PlayFirst and later at Glu Mobile.
Now he’s shifting his focus to virtual reality games. This week he announced he’s joining Resolution Games, the Stockholm, Sweden-based company that Candy Crush Saga co-creator Tommy Palm runs. Hall has joined as a senior creative director and said he’s happy to work on “de-stressors,” or games that reduce daily stress.
Hall’s endorsement comes at an interesting time for VR, which has struggled to find its footing. Initial sales are far below what advocates hoped for around 2014, but passion for VR abides, and headsets such as the Oculus Quest have been selling out during the pandemic.
See the full story here: https://venturebeat.com/2020/06/27/tom-hall-interview-why-doom-co-creator-is-making-virtual-reality-games/
Stuart Russell on AI Governance – AI Futures Podcast (S1E1)
In this episode, Stuart explores some of the correlates between AI governance and the governance and control of other technologies – including electricity and nuclear materials. He also explores some of his thoughts about international AI transparency and governance.
Interview Questions
(3:15) You believe that at some point we’ll need global governance of AI – can you summarize your position on this topic?
(8:15) Are there existing standards, procedures, or laws for other technologies that might apply well to AI as well?
(13:00) Do you think that early AI governance progress will begin with individual, discrete governance instances (say, around using AI to market to children)?
(19:10) Do you think that AI principles can be the right starting point for drilling down to more concrete governance?
(22:10) How can we develop an international AI values consensus with nations that don’t share values such as individualism, privacy, freedom of speech, etc?
(27:00) What does near-term international AI convening look like? How could nations begin to come together?
(33:30) Do you believe that the shared prosperity from beneficial strong AI would relieve the tensions towards violent international competition?
(38:00) What is the role for the private sector in helping to bring us closer to a better AI future?
(41:15) Do you believe that the “new normal” for AI governance and ethics in industries (like insurance or banking) can set a precedent for the bigger picture of AI ethics and governance?
Every Saturday from June 27th to September 12th, we’ll be covering a new AI Futures episode about AI governance.
The podcast can be found here: https://emerj.com/ai-podcast-interviews/stuart-russell-ai-governance/
A reflection on artificial intelligence singularity
Should you feel bad about pulling the plug on a robot or switch off an artificial intelligence algorithm? Not for the moment. But how about when our computers become as smart—or smarter—than us?
Jalsenjak takes us through the philosophical anthropological view of life and how it applies to AI systems that can evolve through their own manipulations. He argues that “thinking machines” will emerge when AI develops its own version of “life,” and leaves us with some food for thought about the more obscure and vague aspects of the future of artificial intelligence.
This is not a discussion about how and when we’ll reach AGI. ... Jalsenjack rather speculates of how the identity of AI (and humans) will be defined when we actually get there, whether it be tomorrow or in a century.
According to philosophical anthropology, the first signs of life take shape when organisms develop toward a purpose, which is present in today’s goal-oriented AI. The fact that the AI is not “aware” of its goal and mindlessly crunches numbers toward reaching it seems to be irrelevant, Jalsenjak says, because we consider plants and trees as being alive even though they too do not have that sense of awareness.
Another key factor for being considered alive is a being’s ability to repair and improve itself, to the degree that its organism allows.
An ideal self-improving AI, however, would be one that could create totally new algorithms that would bring fundamental improvements. This is called “recursive self-improvement” and would lead to an endless and accelerating cycle of ever-smarter AI.
Recursive self-improvement, however, will give AI the “possibility to replace the algorithm that is being used altogether,” Jalsenjak notes. “This last point is what is needed for the singularity to occur.”
See the full story here: https://bdtechtalks.com/2020/06/29/artificial-intelligence-singularity/
Robot with I.A Erica will be the first to star in a movie
Erica will become the first robot with the use of artificial intelligence to have a leading role in a movie as if it were a real actor.
This robot was created practically to act like a real actor and finally he will have his chance to demonstrate everything he is capable of in the seventh art.
See the full story here: https://www.explica.co/robot-with-i-a-erica-will-be-the-first-to-star-in-a-movie/
VR System Hacks Your Nose to Turn Smells Into Temperatures
By using very specific chemicals to access the trigeminal nerve in your nose, they can make you feel hot and cold through smells without realizing you’re smelling anything at all.
iQIYI Announces Winners of its iCartoonFace Challenge, the First Large-scale Cartoon Character Recognition Competition in China
The team from Southeast University won the Challenge with a MAP (Mean Average Precision) of 0.9291 in the cartoon character detection track and the teams from Zhejiang University and Sun Yat-sen University jointly claimed the title of Best Algorithm in the cartoon character recognition track with a Rank1 accuracy of 92.4697%.
While human facial recognition technology is becoming increasingly sophisticated and mature, cartoon character recognition is still in its infancy. Cartoon characters are far more complicated and diverse than human faces in terms of both color and texture. Other challenges include the fact that cartoon characters from the same content or universe could be highly similar in appearance, but characters from different universes may vastly contrast. Due to the complexity of this dataset, cartoon character recognition poses far more challenges than human facial recognition.
Tech Company Offers Free Augmented Reality Fireworks for July 4
The app, TotalAR, puts dazzling displays in the sky through a smartphone or tablet, giving the viewer the opportunity to see fireworks as if standing right under them.
The 12-minute fireworks display will be available to view between 9 p.m. and midnight local time on July 4. Watch it once or multiple times before midnight.
Communities across the country are promoting this virtual display as an alternative to in-person fireworks events.
Available now on TotalAR are other augmented reality experiences - all at no cost and free of ads. Experience a rainbow, floating lanterns, a hot air balloon launch, and a UFO landing all from the comfort of your home or your favorite social-distancing location.
See the full story here: https://www.prnewswire.com/news-releases/tech-company-offers-free-augmented-reality-fireworks-for-july-4-301084106.html
Apple patents reveal ambitions
A review of Apple’s AAPL, -1.47% patent library reveals a deep focus on wearables, health, augmented reality and autonomy, according to Bank of America analyst Wamsi Mohan.
Apple has applied for patents for Apple Watch technology that would measure dyskinesia and tremor symptoms, and it’s been granted one for blood-pressure detection. The company also seems interested in technology like a mat that could measure vital signs and track sleep, augmenting the Apple Watch and other existing iOS devices.
Outside of health, Apple has received a patent for an Apple Watch band that could take photos and videos through an optical sensor.
Apple’s patents also show wide-ranging ambitions for augmented and virtual reality. The company has applied for patents focused on AR navigation systems and AR technology for virtual meetings, both features that could come to a designated AR headset. In addition, Apple has received a patent for an AR device that could warp images so that people who are partially visually impaired could see more clearly.
In autonomous driving, Apple has sought a patent for sensor technology that would gauge stress levels through measures of heartbeat, body temperate and eye movement in order to increase comfort in an autonomous vehicle. The company has been granted a patent for a system that would let drivers give autonomous cars direction about where to park and even use hand gestures to instruct cars to park in a certain lot.
See the full story here: https://www.marketwatch.com/story/apple-patents-reveal-ambitions-in-augmented-reality-and-more-2020-06-25
Pages
- About Philip Lelyveld
- Mark and Addie Lelyveld Biographies
- Presentations and articles
- Trustworthy AI – A Market-Driven approach
- Tufts Alumni Bio