Adobe going to use AI for detecting facial manipulation in images

Artificial Intelligence Can Now Detect Face Which Is Edited

Adobe's prototype AI tool automatically spots Photoshopped faces

By the end of the training, the tool was able to identify manipulated images up to 99% of the time, as compared to the 53% identification rate of humans. The experiment - trying to learn how to recognize manipulation of photos, video, audio and documents.

The researchers at Adobe and UC Berkeley have published their work in a new paper titled, "Detecting Photoshopped Faces by Scripting Photoshop", explaining how the new method can figure out if Photoshop's Face Aware Liquify feature was used.

The team working on the project trained a convolutional neural network (CNN) to spot changes in images made with Photoshop's Face Away Liquify feature, which was created to change people's eyes, mouth and other facial features. "The feature's effects can be delicate which made it an intriguing test case for detecting both drastic and subtle alterations to faces", said Adobe.

Former NFL Running Back Rod Smart Missing, Police Seek Assistance
Police in SC have put out a missing person advisory for former Philadelphia Eagles and Carolina Panthers running back Rod Smart. The Lancaster County Sheriff's Office is asking anyone with information about Smart to call its office.


But with the threat of image and video fakery quickly reaching epic proportions (think deepfakes and face swapping), fake content, images and videos are becoming a major issue. Of those images, some were chosen to train the CNN, with additional help with an artist's touch.

And when it comes to spotting manipulated images and adjusted faces, Adobe says this is factual the starting. "For this approach to be useful, it should be able to perform significantly better than the human eye at identifying edited faces", said Adobe.

It showed the altered and unaltered images to some human volunteers to see how good or bad they are to identify which photo was fake.

Suppression de la taxe d'habitation: le calendrier définitif
Cela représenterait, selon les calculs des équipes de Bercy, un gain moyen de 804 euros par foyer et par an dans les deux collectivités réunies.


UC Berkeley researcher Professor Alexi A. Efros explains that detecting image fakery may seem impossible because there are many elements to facial geometry.

It was even possible for the tool to revert images which had been altered back to how they had looked before.

Editing the photo is quite a common phenomenon now with the help of the Photoshop software and the leader of this software in non-other than Adobe.

MotoGP: Marquez gagne en Catalogne, 1er podium pour Quartararo
C'est l'Italien Andrea Dovizioso (Ducati), le dauphin de Marquez pour 12 points, qui hérite de la sixième place sur la grille. La course reprendra ses droits le 30 juin à Assen, aux Pays-Bas, temple de la vitesse moto.


"That is the most important step in with the ability to detect certain forms of image modifying, and the undo capacity works strangely well". "But we live in a world where it's becoming harder to trust the digital information we consume, and I look forward to further exploring this area of research". "Beyond technologies like this, the most effective defense will probably be a sophisticated public who recognize that content may also be manipulated, often to please them, however sometimes to mislead them as well", mentioned Gavin Miller, Head of Analysis, Adobe.

Latest News