Tutorial and stats to build a simple machine learning classifier using Tensorflow + Keras
I am HEAVILY copying most if not all of the content from https://medium.com/@seixaslipe/building-a-simpsons-classifier-with-deep-learning-in-keras-36a47fe17f79
This is just my attempt to try this locally on my own system, and also do some benchmarks
This is based on these posts: https://medium.com/alex-attia-blog/the-simpsons-character-recognition-using-keras-d8e1796eae36
and has code from here https://github.com/alexattia/SimpsonRecognition.git
https://www.pyimagesearch.com/2017/09/25/configuring-ubuntu-for-deep-learning-with-python/
sudo pip install virtualenv virtualenvwrapper
mkvirtualenv simpsonsKeras
Download the data set from here: https://www.kaggle.com/alexattia/the-simpsons-characters-dataset
mkdir rawImageData cd rawImageData unzip ~/Downloads/the-simpsons-character-dataset.zip
annotation.txt characters_illustration.png kaggle_simpson_testset number_pic_char.csv simpsons_dataset.tar.gz weights.best.hdf5
tar -zxvf simpsons_data.tar.gz
simpsonsZipData or similar folder
cd web ln -s ../RawImageData/ bower install
python -m SimpleHTTPServer
mkvirtualenv KerasSimpsons -r requirements.txt
We want these plots: https://medium.com/alex-attia-blog/the-simpsons-character-recognition-using-keras-d8e1796eae36
https://www.pyimagesearch.com/2017/09/25/configuring-ubuntu-for-deep-learning-with-python/