Communications of the ACM
AI Removes Unwanted Objects From Photos
Virginia Polytechnic Institute and State University researchers developed an algorithm that uses artificial intelligence to digitally remove unwanted objects from photos.
Credit: Xue, Tianfan, Michael Rubinstein, Ce Liu, and William T. Freeman
Researchers at Virginia Polytechnic Institute and State University developed an algorithm that uses artificial intelligence to digitally remove unwanted objects in photos.
The neural network analyzes several frames in a motion photo taken by a smartphone, identifies various objects in the image, and calculates the distance to each object.
It then separates the objects into different layers, removing the foreground layer to provide an unobstructed view of the objects behind.
The neural network automatically identifies distracting objects, including reflections, in the foreground of the image during the process of learning.
The algorithm works faster than prior methods, taking just three minutes to decipher a 1,296-by-864-pixel image.
From New Scientist
View Full Article - May Require Paid Subscription
Abstracts Copyright © 2020 SmithBucklin, Washington, DC, USA
No entries found