A neural network is created to improve the quality of GIF files 2017-12-21 11:27:00 / TECHNOLOGİCAL İNNOVATİONS

Hosting Gfycat has created a neural network that improves the quality of blurry gif images using machine learning. The program searches the Internet for the original version of the video from which the hypha was taken and analyzes it to find out which part has turned into an animation. Then the neural network creates a higher quality gif and pours it into the network, thus promoting the dissemination of high-quality images.

In addition, the artificial intelligence Gfycat enhances the quality of the font on the animated image. Another function of the neural network is the recognition of faces of famous people on animations and automatic tagging with names to such images. According to the director general of Gfycat Richard Rabbat, by now the algorithm has processed about two million gifs.

    Copyright © AMEA İnformasiya Texnologiyaları İnstitutu, 2017