How Computer Vision Is Transforming MediaComputer Vision
A New Way to Watch the GameNew York Times
Better Smartphones Through Deep LearningAdAge
The Well-Read RobotImaging and Vision Europe
Scanning books to digitize them is a long tedious process because someone has to actually turn those pages and make sure they’re positioned properly to avoid any distortions. Is it any wonder your favorite out-of-print books still aren’t available in digital form? Good news: You may not have to scour through used bookstores or library bookshelves to read those tomes thanks to a group of engineers in Greece. Its company, Bookscanner, has developed a new glass-enclosed, computer vision-enabled robot that is able to scan each and every page of a book without doing any damage. Distortions in the images of the scanned pages are evened out with the help of neural networks.
Google MoonU of T News
Some of space’s biggest secrets can be revealed by something as simple as counting craters on the surface of the moon, but getting humans to do it accurately is a challenge to say the least, especially when you consider that smaller craters within bigger craters that reveal just how old a surface formation is. Researchers at the University of Toronto are now using convolutional neural networks to take on the task, and it’s working quite well, with the models able to identify about 6,000 additional craters on the moon so far. The technology will also be applied to similarly cratered surfaces on Mercury.