A Video Summarization Method for Movie Trailer-Genre Classification Based on Emotion Analysis
摘要
We live in an information world where visual data undergo exponential growth within a very short time window. With diverging content diversity, we simply have no capacity to keep track of those data. While short video platforms (such as TikTok™ or YouTube Shorts™) can helped users viewing relevant videos within the shortest time possible, those videos might have misleading information, primarily if it is derived from long videos. Here, we analyzed several short videos (in terms of movie trailers) from YouTube and established a correlation between one movie trailer and the classified movie genre based on the emotion found in the trailer. This paper contributes to (1) an efficient framework to process the movie trailer and (2) a correlation analysis between the movie trailer and movie genre. We found that every movie genre can be represented by two unique emotions.