For movie lovers, the movie poster makes up the first impression about the movie contents and its genre.
Humans can grasp the cues like color, expressions on the faces of actors etc to quickly determine the genre (horror, comedy, animation etc) of a movie. Though it might seem simple, it involves millions of concurrent processes occurring in a less than a jiffy! If humans are able to inherently predict genre of a movie by a glance at its poster, can't we expect the same from a machine ?
It has been studied that the human brain tends to make intuitions based on the color characteristics, local texture based features and structural cues of posters. Hence, the posters possess some characteristics which could be utilized for genre classification
In this competition, your goal is to predict the genres of movies from their posters.
- Participants must submit a kaggle kernel which reproduces their selected submission and make the kernel public after the contest end to be eligible for evaluation and prizes.
- Please have a look at the competition rules. All rules are to be followed strictly. Any use of unfair means will result in immediate disqualification from the competition. The competition was organized by CyberLabs, the official Coding and Cyber Club of IIT (ISM) Dhanbad, India.
My work includes:
- Basic Fast A.I implementation
- Model[gluon-nets,resnexts,resnets,efficientnets]
- Optimizer[Adam]
- Loss Function[Cross-Entropy]