Adobe use machine learning to facilitate the detection of fake pictures

This topic Adobe use machine learning to facilitate the detection of fake photos appeared on Engadget.

Getting worried experts around the world on new tools to assess the artificial intelligence which facilitates than ever to the process of editing photos and videos private when you publish and share content via social networking sites, without verifying the authenticity of this content.

And is the development of some of these tools by Adobe, but the company is also working to use the technique of the interview and that by searching for how to use machine learning in the identification of images edited automatically.

Where was the clarification of the latest work of the company, and presented this month at the Conference on CVPR, how to solve forensic digital performed by humans, by machines or artificial intelligence technology in much less time.

And the company spokesman said that this project was a research project at an early stage, but the company wishes in the future to play a role in the development of technology that helps in monitoring the health of digital media and verified.

It is worth noting that Adobe did not unlock by software designed to detect fake pictures. But, the company refers to its work with law enforcement using digital forensics to help find missing children, for example. And as evidence of its position in charge towards their own technology.

And show research paper new how can you use machine learning for three common types of image manipulation: profit, where it is combine two parts of different images. Reproduction, where the copy and paste things within the image. And removal, when you are editing something in the image perfectly.

And for this kind of manipulation, looking the experts of digital forensics are usually clues in the layers of the hidden image. When you are making these types of adjustments, they leave behind digital traces, such as the development of the random variations in color and brightness created by the photo sensors. And that appear when you collect two different images, for example, or copy something from one part of the picture and paste it.

And as is the case with many machine learning systems other, it has been teach Adobe software through the use of a large collection of images liberated. And through it, learn the program how to determine the common patterns that indicate manipulation in the photo.

Source

This topic Adobe use machine learning to facilitate the detection of fake photos appeared on Engadget.

0 Comments on “Adobe use machine learning to facilitate the detection of fake pictures”

Leave a Reply

Your email address will not be published. Required fields are marked *