Adobe’s new AI software can identify a real image from its fake counterparts

Photoshop is one of that software which is used around the globe to transform images. The software is widely used by creative artists to develop their artwork. However, sometimes it becomes hard for the users to identify a real image of its fake counterparts which are also developed using Photoshop.

It is a well-known fact that expert Photoshoppers can easily develop images which could be hard to identify among its real photos. In order to tackle the situation, Adobe is now training an AI which could carry out the task of identifying the real images among the fake counterparts. The AI will closely study the patterns of noise across the image and spot the abnormalities in the images to figure out the fake or modified image.

This feature will come in handy in the areas of law where the images are being used as forensic evidence. The AI's judging capabilities are still being questioned; However, Adobe has also shared a few examples to illustrate the AI's capabilities.

The major concern which remains to be answered is that how accurately the AI will be able to identify the fake images when the image quality of an image is poor or not much clear which majorly depends on the images that are used for training the digital brain of the AI. It is also not surprising that AI is making such progress as we have also earlier seen the AI "Deepfake". However, this implies that Adobe is trying to fix the malicious use of its power-packed software.

Vlad Morariu, a senior research scientist at Adobe and lead author of the paper on Computer Vision said that apart from identifying the odd spots in an image, the new AI method can also detect and differentiate between the various types of "tampering techniques". The major battle which Adobe will have to fight is the AI's ability to identify fake images created using other AI's. Recently, Adobe has also introduced a whole set of features in Photoshop CC 2018, which allows a user to create some powerful images.