Movie genres and beyond: An analytical survey of classification techniques

In today's world with a plethora of options available on movie content, it is very important to have models which could classify these large contents based on their genres. Movies are the most widely entertaining source all over the world. Every person has his own taste of movies, some feel action movies are better, some would say romantic. So the online platforms must have the collection of movies according to a particular category i.e. based on genres. This paper has proposed an extensive and exhaustive survey for various models and techniques, which could be applied to solve the challenges involved in performing the genre based classification. This paper summarizes and compares all those techniques by classifying them into four broad domains -through text mining, poster detection, audio and video features of the movies. These domains included various classifying techniques: Bi-directional LSTM, gram layer, scene categorization and CNN respectively. With such an extensive amount of work going in the field in the last not many years, this review paper will help specialists in the field of movie genre classification.

I. Introduction

Video based content assume a major function in satisfying individuals' delight needs. On account of Internet innovation headways, content consumers approach a remarkable measure of motion pictures from different online administrations.//With increasing number of users using the internet in India and around the world have shot up multiple folds in recent past, thanks to free internet and lower rate/cheap internet packs. Because of this there has been a sudden increase in number of people engaging in consumption of online entertainment channels having web shows, web series, movies, daily soaps etc. This has made the need for programmed content-driven moving-picture show suggestions of genres before a user starts watching it.