A Deep Learning approach of video compression

This project tackles the issue of video compression with a Deep Learning approach. More precisely, it focuses on the case of slideshows broadcasts (which have become popular with the development of Internet, and even more popular this last year with the sanitary crisis). With Deep Learning tools, we detect the different elements of a slide to process them differently in function of their type (text, image, background, charts ...) minimizing the quantity ofinformation to send.