This project is about recognizing main pokemon type(water, fire, grass). Convolutional neural network model used is Alexnet. Some visulizations potions and data augmentation are done before CNN model.
The choice of number of epochs is based on the experiment output(accuracy stops increasing). Hyperparameter tuning includes changes in batch size, different model optimizer and learning rate, droupout layers, etc.
Final ouptput is considerable since limited datas are provided to the model. dataset link: https://www.kaggle.com/vishalsubbiah/pokemon-images-and-types