Showing posts with label believing. Show all posts
Showing posts with label believing. Show all posts

Wednesday, June 16, 2021

Mind Behind the Eyes


Just human things.

An averted glance gives a glimpse of the mind behind the eyes
Aug 2020, phys.org

The amount of stuff our brain does without our even thinking about it, is unfathomable:
In almost all cases, people instinctually follow the gaze of another. But psychologists found an exception in the socially awkward situation where a person caught staring averts their eyes: A third-party observer does not reflexively follow their gaze. (Which means they know, albeit unconsciously.)

The researchers conclude that the brain tells the observer that there is no significance to the location where the embarrassed party has turned their attention.

"The brain is a lot smarter than we thought," said Yale's Brian Scholl, professor of psychology and senior author of the paper.

via: Clara Colombatto el al., "Gaze deflection reveals how gaze cueing is tuned to extract the mind behind the eyes," PNAS (2020). www.pnas.org/cgi/doi/10.1073/pnas.2010841117

Why the brain is programmed to see faces in everyday objects
Aug 2020, phys.org
They tested this using the process known as "sensory adaptation," a kind of visual illusion where one's perception is affected by what has recently been seen: 

"If you are repeatedly shown pictures of faces that are looking towards your left, for example, your perception will actually change over time so that the faces will appear to be looking more rightwards than they really are," says Dr. Palmer.

via: Colin J. Palmer et al. Face Pareidolia Recruits Mechanisms for Detecting Human Social Attention, Psychological Science (2020). DOI: 10.1177/0956797620924814
^Somewhat related to right angle bias?


Study of ancient Mayan facial expressions suggests some are universal
Aug 2020 phys.org

There's a good paradox about whether human facial expressions are universal. Would a person from across the world understand you just by your facial expression? The paradox is that in order to do this study, a human observer has to be involved, which means they're biased, because they are part of a culture that has programmed their brain since birth to see things a certain way. But not robots. They aren't born at all!

Just kidding, they used English-speaking mechanical turks to label photos of Mayan sculptures. The turks agreed on the photos.

via: Alan S. Cowen et al. Universal facial expressions uncovered in art of the ancient Americas: A computational approach, Science Advances (2020). DOI: 10.1126/sciadv.abb1005

Totally unrelated image credit: bushmeat, via tengwood.org


Wednesday, January 13, 2021

Face Rec and Image Tech


Deepfake used to attack activist couple shows new disinformation frontier
Experts in deceptive imagery used state-of-the-art forensic analysis programs to determine that Taylor’s profile photo is a hyper-realistic forgery - a “deepfake.”
Something about the Ventriloquist's Dummy and the dangers of impersonating an influential leader.



Image cloaking tool thwarts facial recognition programs
Aug 2020. phys.org

Reverse camoufaluge? Fill the pool with dirty data; it's a simple approach. This technique is named Fawkes after the V for Vendetta guy, also known as the Anonymous Mask from the online shadow activist group circa 2010.

"What we are doing is using the cloaked photo in essence like a Trojan Horse, to corrupt unauthorized models to learn the wrong thing about what makes you look like you and not someone else," Fawkes co-creator Ben Zhao, a computer science professor at the University of Chicago, said.

"Our original goal was to serve as a preventative measure for Internet users to inoculate themselves against the possibility of some third-party, unauthorized model," the team recently stated in a FAQ sheet.


Deepfake detection tool unveiled by Microsoft
Sep 2020, BBC News
"The only really widespread use we've seen so far is in non-consensual pornography against women," commented Nina Schick, author of the book Deep Fakes and the Infocalypse.

"But synthetic media is expected to become ubiquitous in about three to five years, so we need to develop these tools going forward

Synthetic media and fingerprinted news - Microsoft has teamed up with the BBC, among other media organisations, to support Project Origin, an initiative to "mark" online content in a way that makes it possible to spot automatically any manipulation of the material.

New photon-counting camera captures 3-D images with record speed and resolution
Apr 2020, phys.org

24,000 fps:
The camera's speed makes it possible to measure the time a photon hits the sensor very precisely. This information can be used to calculate how long it takes individual photons to travel the distance from a source to the camera, known as time-of-flight. Combining time-of-flight information with the ability to capture a million pixels simultaneously enables extremely high-speed reconstruction of 3-D images.