An Multi-Dimensional Video Reverse Search Engine using Computer Vision and Machine Learning

Online media has become a mainstream of current society. With the rapid development of video data, how to acquire desired information from certain provided media is an urgent problem nowadays. The focus of this paper is to analyse a sufficient algorithm to address the issue of dynamic complex movie classification. This paper briefly demonstrates three major methods to acquire data and information from movies, including image classification, object detection, and audio classification. Its purpose is to allow the computer to analyse the content inside of each movie and understand video content. Movie classification has high research and application value. By implementing described methods, finding the most efficient methods to classify movies is the purpose of this paper. It is foreseeable that certain methods may have advantages over others when the clips are more special than others in some way, such as the audio has several significant peaks and the video has more content than others. This research aims to find a middle ground between accuracy and efficiency to optimize the outcome.