The technology to clean up poor photos has been around for a while now. This technology is vital, because often photos will prove pixilated. Identities of person’s in the photo can be unclear due to missing features and other important clarity issues. Now, thanks to exciting new developments, the restoration of such photos may be entirely achievable even without a second, accurate photo to compare it with. A new type of implemented, machine-specific learning has now trained AI systems devised to restore such photos to do so even in the absence of clear data. Now, using only corrupted photos, the systems can still recreate an entirely accurate photo, on par with that generated with the use of clean data. This technology can be utilized for text and data and has particularly exciting ramifications for the field of medicine.

Key Takeaways:

• The technology to repair corrupted photos by comparing them to clean data has been around for a while now.
• A new machine-specific learning now has AI systems restoring corrupted photos that are on par with those using clean data, without having used such data.
• The technology saves time and should have vital strategic use in the medical field, where clean images, such as MRIs are of the upmost importance.

“The AI does this by utilizing a neural network that’s been trained using corrupt photos. It doesn’t need a clean image, but it does need to observe the source image twice.”

Read more: https://www.engadget.com/2018/07/10/nvidia-ai-fix-bad-photos-deep-learning/