A smallerVGGNet-Model (pyimagesearch.com) was trained to classify five diefferent pokemons (pikachu, bulbasaur, charmander, mewtwo, squirtle). You can either start the GUI or run the classify.py script in the CLI.
The python script train.py creates a trained model, a lb.pickle file and polts the train results.
python train.py --dataset [path to dataset] --model [name of model to create] --labelbin [name of lb.pickel to create]
Example:
python train.py --dataset dataset --model pokedex.model --labelbin lb.pickle
The script classify.py classifies an example image and shows the image with the result.
python classify.py --model [path to trained model] --labelbin [path to pickle file] --image [path to image]
Example:
python classify.py --model pokedex.model --labelbin lb.pickle --image examples\pikachu.png
The script index.py starts the GUI.
This project was built following along this tutorial: https://www.pyimagesearch.com/2018/04/16/keras-and-convolutional-neural-networks-cnns/