Pokedex_project_V1.672 at Le Wagon

This project was done at le Wagon in two weeks, during the Data Science Bootcamp. With this project you can classify by types and names the 151 first pocket monster (A.K.A Pokémon), thanks to a CNN model. But that's not it ! You can also genrate new one thanks to a GAN model.

To test the app, go on this website : https://pokemon-generator-1672.streamlit.app/

Installation

First let's clone the repository :

git clone https://github.com/Just-PH/lewagon-pokedex-gan.git

The run the installation :

cd backend
make start

⚠️ Then you need to modify in file /backend/.env the variable WHO with your name ⚠️

Test

Still in /backend To test the both predictions function on all images :

make run_test

If you only want to test for the types :

make run_test_15

If you only want to test for the names :

make run_test_150

Predictions

⚠️ PUT any images you want to test in the repository all_prediction_images/images ⚠️

To predict them :

make run_pred

If you only want to predict for the types :

make run_pred_15

If you only want to predict for the names :

make run_pred_150

Generation

To generate this kind of images :

You can use this command :

make run_generate

The image will be in the repository named output_gan

Api

To run locally the api :

make run_api_local