Photoshop is arguably the most powerful tool at a photographer’s disposal. And like many powerful tools, it can be used for either good or harm. The good news is that technology has recently been developed to counter nefarious and sneaky image manipulations.
Adobe and the University of California Berkeley have teamed up to develop new software that detects facial manipulations that have been implemented on images by Photoshop’s Face Aware Liquify tool.
In a study, both humans and AI and computers were tested on their ability to detect altered facial features. Not surprisingly, AI was the overwhelming winner. But what is surprising is that AI not only could detect the facial “fakes” with 99% accuracy, it could even revert those images to their original state.
In the age of Photoshop and of facial manipulation apps, the lines between reality and fantasy have become blurred. Photographers have been caught and ousted for cheating on contests with composites, while photojournalists have been reprimanded for manipulating seemingly accurate images. And rightfully so…Read more
Using Photoshop and other image manipulation software to tweak faces in photos has become common practice, but it’s not always made clear when it’s been done. Berkeley and Adobe researchers have created a tool that not only can tell when a face has been Photoshopped, but can suggest how to undo it.
Right off the bat it must be noted that this project applies only to Photoshop manipulations, and in particular those made with the “Face Aware Liquify” feature, which allows for both subtle and major adjustments to many facial features. A universal detection tool is a long way off, but this is a start.
The researchers (among them Alexei Efros, who just appeared at our AI+Robotics event) began from the assumption that a great deal of image manipulation is performed with popular tools like Adobe’s, and as such a good place to start would be looking specifically at the manipulations possible in those tools.
They set up a script to take portrait photos and manipulate them slightly in various ways: move the eyes a bit and emphasize the smile, narrow the cheeks and nose, things like that. They then fed the originals and warped versions to the machine learning model en masse, with the hopes that it would learn to tell them apart.
Learn it did, and quite well. When humans were presented with images and asked which had been manipulated, they performed only slightly better than chance. But the trained neural network identified the manipulated images 99 percent of the time.
What is it seeing? Probably tiny patterns in the optical flow of the image that humans can’t really perceive. And those same little patterns also suggest to it what exact manipulations have been made, letting it suggest an “undo” of the manipulations even having never seen the original.
Since it’s limited to just faces tweaked by this Photoshop tool, don’t expect this research to form any significant barrier against the forces of evil lawlessly tweaking faces left and right out there. But this is just one of many small starts in the growing field of digital forensics.
“We live in a world where it’s becoming harder to trust the digital information we consume,” said Adobe’s Richard Zhang, who worked on the project, “and I look forward to further exploring this area of research.”…Read More
In a world where technology is often used to manipulate photos to make them appear real, it sometimes is difficult to determine if a photo is edited or not — but not for long as a team of researchers from Adobe and UC Berkeley have developed an AI to do just that.
In a new paper titled “Detecting Photoshopped Faces by Scripting Photoshop,” the researchers Sheng-Yu Wang, Oliver Wang, Andrew Owens, Richard Zhang and Alexei A. Efros said they were able to create a new AI that is capable of detecting images that were edited or warped using Photoshop.
How they did it
The team explained that they were able to create the AI by scripting Photoshop, and by feeding the AI fake images until it was able to detect the fake from the real photos. PetaPixel reported that the researchers used thousands of photos scraped from the internet. These photos were either edited automatically using a Photoshop script, or edited by a human artist.
They then compared the number of successful photoshopped image detections both people and the AI can do.
“We started by showing image pairs (an original and an alteration) to people who knew that one of the faces was altered,” said researcher Oliver Wang. “For this approach to be useful, [the AI] should be able to perform significantly better than the human eye at identifying edited faces.”
The researchers proudly said that the AI is better than people at detecting edited from non-edited images. Humans were able to identify edited photos about slightly more than half of the time (53 percent). The AI, on the other hand, successfully detected edited images almost all of the time (99 percent).
More than just knowing if an image is photoshopped, the AI is also capable of determining what part of the image was edited or warped, and is also capable of “undoing” the edit to return the image to how it originally looked like…Read More
Kendrick Lamar isn’t the only person that wants us to be “Humble” on Instagram with unedited photos. A new AI-based piece of software can detect a photoshopped face and even reveal what the picture was prior to the edits being applied.
Researcher for Adobe Oliver Wang said of the process, “We started by showing image pairs (an original and an alteration) to people who knew that one of the faces was altered…For this approach to be useful, it should be able to perform significantly better than the human eye at identifying edited faces.”
The AI performed way better than any human could at the task and could even reverse the edits performed on the photo. With a detection rate accuracy of 99%, the AI’s ability to reverse engineer an edited photo bodes well for the research project’s hope of using things like this to verify the accuracy of photos and videos in the future. As editing and manipulation become increasingly sophisticated, the greater the need for software capable of detecting edits and changes…Read more