NVIDIA has unveiled an impressive new AI that can automatically remove noise, grain, and even watermarks from photos. What’s even more impressive is the fact that it’s all accomplished without studying any examples of “clean photos” — all the AI needs are examples of corrupted images.
The AI, developed by researchers at NVIDIA, MIT, and Aalto University, is different than how other state-of-the-art noise-reduction AI systems work.
“Recent deep learning work in the field has focused on training a neural network to restore images by showing example pairs of noisy and clean images,” NVIDIA writes. “The AI then learns how to make up the difference. This method differs because it only requires two input images with the noise or grain.
“Without ever being shown what a noise-free image looks like, this AI can remove artifacts, noise, grain, and automatically enhance your photos.”
After training the AI on 50,000 photos — you can read the technical paper here — the AI was able to produce extremely impressive results.
Here are some before-and-after examples of noise and grain removal done by the AI:
In addition to noise and grain, the AI can also automatically remove artifacts… including random watermarks splashed all across the photo:
Here’s a video with a few more before-and-after examples:
NVIDIA says there are several real-world applications for this AI, as sometimes there simply aren’t any actual “clean” examples to train AI with. Examples include low-light photography in certain fields (e.g. astrophotography) in which low-quality images are all we have.