The goal of the project was to develop a web scraper to automatically find video inputs on Google and Instagram to create deep fake videos.
In this project, a novel method of image identification to enhance the quality of the deep fake videos was developed and deployed. By using this method, the time for preprocessing could be significantly reduced.
In addition, a proof of concept of the whole process with an implementation in Python was given. Deep fakes are of particular relevance in the media sector, for example to rejuvenate actors.