Built using fast.ai library.
Memes, Quotes, Film-Poster these three almost have similar characteristics and appear the same in many cases. So, I thought of making a try make the computer learn and detect and differenciate these three things.
This is model is trained with a dataset which built by my own from google images.
The dataset contains 2470 images which include memes, quotes and film posters for training and the validation set consists of 617 images.
The model is rained using a pretrained model called the resent34.
Appropriatly using those I managed to get an accuracy of 93.5% The model perfectly predicts the memes,quotes and film posters.