In an era where artificial intelligence (AI) image creation is within everyone’s reach, the ability to detect fake images – particularly deepfakes of people – is becoming increasingly important.
What if you could tell just by looking someone in the eye?
That’s the compelling conclusion of new research shared at the Royal Astronomical Society’s national astronomy meeting in Hull, which suggests that AI-generated fakes can be spotted by analyzing human eyes in the same way astronomers study images of galaxies.
Much of the work of Adejumoke Owolabi, a master’s student at the University of Hull, focuses on the reflection in a person’s eyeballs.
If the reflections match, the image is probably of a real human. If they don’t, it’s probably a deepfake.
A series of deepfake eyes showing inconsistent reflections in each eye.
Adejumoke Owolabi
“The reflections in the eyeballs are consistent for the real person, but incorrect (from a physical point of view) for the fake person,” said Kevin Pimbblet, professor of astrophysics and director of the Centre of Excellence for Data Science, Artificial Intelligence and Modelling at the University of Hull.
The researchers analyzed light reflections on people’s eyeballs in real and AI-generated images. They then used methods commonly used in astronomy to quantify the reflections and check for consistency between the reflections from the left and right eyeballs.
Fake images often lack consistency in the reflections between each eye, whereas real images usually show the same reflections in both eyes.
“To measure the shape of galaxies, we analyze whether they are compact at the center, whether they are symmetrical and whether they are smooth. We analyze the distribution of light,” explains Professor Pimbblet.
“We detect reflections in an automated manner and run their morphological characteristics via CAS (concentration, asymmetry, softness) and Gini indices to compare the similarity between left and right eyeballs.
“The results show that deepfakes have some differences between the two.”
A series of real eyes showing largely consistent reflections in both eyes.
Adejumoke Owolabi
The Gini coefficient is commonly used to measure how the light in an image of a galaxy is distributed among its pixels. This measurement is made by ranking the pixels that make up a galaxy image in order of increasing flux, and then comparing the result to what would be expected from a perfectly uniform flux distribution.
A Gini value of 0 corresponds to a galaxy in which light is evenly distributed across all pixels in the image, while a Gini value of 1 corresponds to a galaxy with all the light concentrated in a single pixel.
The team also tested CAS parameters, a tool originally developed by astronomers to measure the light distribution of galaxies to determine their morphology, but found that it did not effectively predict false eyes.
“It is important to note that this is not a silver bullet for detecting fake images,” added Professor Pimbblet.
“There are false positives and false negatives, it’s not going to catch everything. But this method gives us a baseline, a plan of attack, in the arms race to detect deepfakes.”
Media Contacts
Sam Tonkin
Royal Astronomical Society
+44 (0)7802 877 700
press@ras.ac.uk
Dr. Robert Massey
Royal Astronomical Society
Mobile: +44 (0)7802 877 699
press@ras.ac.uk
Megan Eaves
Royal Astronomical Society
press@ras.ac.uk
Scientific contacts
Kevin Pimbblet
University of Hull
K.Pimbblet@hull.ac.uk
Images and captions
Real image vs fake
Caption: In this image, the person on the left is real, while the person on the right is AI-generated. Their eyeballs are shown below their face. The reflections in the eyeballs are consistent for the real person, but incorrect (from a physical perspective) for the fake person.
Credit: Adejumoke Owolabi
Examples of deepfakes
Caption: A series of fake eyes showing inconsistent reflections in each eye.
Credit: Adejumoke Owolabi
Real eye examples
Caption: A series of real eyes showing largely consistent reflections in both eyes.
Credit: Adejumoke Owolabi
Notes to Editors
The NAM 2024 conference is principally sponsored by the Royal Astronomical Society, the Science and Technology Facilities Council and the University of Hull.
About the Royal Astronomical Society
The Royal Astronomical Society (RAS), founded in 1820, encourages and promotes the study of astronomy, solar system science, geophysics and closely related branches of science.
The RAS organises scientific meetings, publishes international research and review journals, recognises outstanding achievement with medals and prizes, maintains an extensive library, supports education through grants and outreach activities and represents British astronomy nationally and internationally. Its over 4,000 Fellows, a third of whom are based overseas, include research scientists at universities, observatories and laboratories as well as historians of astronomy and others.
The RAS accepts articles for publication in its journals on a peer-review basis, whereby the experts on the editorial boards agree that the article is worthy of consideration. The Society issues press releases on a similar basis, but the organisations and scientists concerned are generally responsible for their content.
Stay up to date with the RAS on XFacebook, LinkedIn and YouTube.
About the Science and Technology Facilities Council
The Science and Technology Facilities Council (STFC) is part of UK Research and Innovation, the UK body that works in partnership with universities, research organisations, businesses, charities and government to create the best possible environment for research and innovation to flourish.
The STFC funds and supports research in particle and nuclear physics, astronomy, gravitational research and astrophysics, and space science. It also manages a network of five national laboratories, including the Rutherford Appleton Laboratory and the Daresbury Laboratory, and supports UK research at a number of international research facilities, including CERN, FERMILAB, the ESO telescopes in Chile and many others.
The STFC Astronomy and Space Science Program supports a wide range of facilities, research groups, and individuals to study some of the highest-priority questions in astrophysics, cosmology, and solar system science.
The STFC’s astronomy and space science programme is supported by grants for research activities, as well as support for technical activities at the STFC’s UK Astronomy Technology Centre and RAL Space at the Rutherford Appleton Laboratory. The STFC also supports UK astronomy through the European Southern Observatory and the Square Kilometre Array Organisation.
For more information visit https://stfc.ukri.org/. Follow STFC on Twitter: @STFC_Matters
About the EA Milne Centre at the University of Hull
The EA Milne Centre for Astrophysics at the University of Hull brings together experts who study the evolution of the structure of the Universe, from stars to galaxies and galaxy clusters, to the largest structures in the cosmos.
The centre uses observations, theories and computational methods in collaboration with international partners. Undergraduate and postgraduate students work alongside staff to understand the wonders of the Universe. Through a series of outreach activities, the centre also aims to share its passion for astronomy and astrophysics with the region and beyond.