Creating Cover Photos (Thumbnail) for Movies and TV Series with Convolutional Neural Network

摘要

The latest version and the use of video streaming platforms is increasing day by day, adding new ones. Competition is also increasing in the development of platforms for expanding film and TV series. The purpose of replicating these platforms is to achieve more, better quality on some platforms, and to keep them on one platform even better. Before that, film and TV series platforms use artificial intelligence algorithms. In this study, the aim is to create more attractive cover photos for the user by finding suitable frames from a movie or TV series and converting these frames into cover/thumbnail images on the platform. It is based on eliminating frames that are useless according to closed eyes, blurred frames or non-face images. Also, deep learning used for labeling images with objects and face’s emotion and identity.