Nvidia and scientists teach AI to remove graininess from photos

Researchers from MIT, Aalto University and Nvidia cleans up the noise from pictures

Scientists have developed an artificial intelligence system that can automatically remove noise, specks, and other distortions from pictures.

The technology, called  Noise2Noise AI, was developed by researchers from Nvidia, Aalto University in Finland, and MIT. The researchers used 50,000 pictures, as well as MRI scans and computer-generated noisy images, to train the system. According to the research paper, the AI can remove enough noise to make images usable again without ever seeing a clean image.

Advertisement - Article continues below

Nvidia Tesla P100 GPUs with the cuDNN-accelerated TensorFlow deep learning framework were used to train up the system.

The technology has been trained to remove noise without needing to understand what a clean image looks like, which until now, such AI work has focused on training a neural network to restore images by showing example pairs of noisy and clean images.

"It is possible to learn to restore signals without ever observing clean ones, at performance sometimes exceeding training using clean exemplars," the researchers said.

They added that the neural network is "on par with state-of-the-art methods that make use of clean examples using precisely the same training methodology, and often without appreciable drawbacks in training time or performance".

The system was tested using three different datasets to validate the neural network.

Advertisement
Advertisement - Article continues below

Scientists said there were several real-world situations where obtaining clean training data is difficult: low-light photography (e.g., astronomical imaging), physically-based image synthesis, and magnetic resonance imaging.

Advertisement - Article continues below

"Our proof-of-concept demonstrations point the way to significant potential benefits in these applications by removing the need for potentially strenuous collection of clean data,"  said the paper's authors. "Of course, there is no free lunch we cannot learn to pick up features that are not there in the input data but this applies equally to training with clean targets." 

Featured Resources

Top 5 challenges of migrating applications to the cloud

Explore how VMware Cloud on AWS helps to address common cloud migration challenges

Download now

3 reasons why now is the time to rethink your network

Changing requirements call for new solutions

Download now

All-flash buyer’s guide

Tips for evaluating Solid-State Arrays

Download now

Enabling enterprise machine and deep learning with intelligent storage

The power of AI can only be realised through efficient and performant delivery of data

Download now
Advertisement

Most Popular

Visit/security/privacy/355155/zoom-kills-facebook-integration-after-data-transfer-backlash
privacy

Zoom kills Facebook integration after data transfer backlash

30 Mar 2020
Visit/security/data-breaches/355173/marriott-hit-by-data-breach-exposing-personal-data-of-52-million
data breaches

Marriott data breach exposes personal data of 5.2 million guests

31 Mar 2020
Visit/security/cyber-crime/355171/fbi-warns-of-zoom-bombing-hackers-amidst-coronavirus-usage-spike
cyber crime

FBI warns of ‘Zoom-bombing’ hackers amid coronavirus usage spike

31 Mar 2020
Visit/data-insights/data-management/355170/oracle-cloud-courses-are-free-during-coronavirus-lockdown
data management

Oracle cloud courses are free during coronavirus lockdown

31 Mar 2020