shrestha-pranav/quickdraw

QuickDraw Image Recognition

Pranav Shrestha (ps2958) Jeffrey Wan (jw3468)

Current Kaggle position: 388 (0.908)

Exploring the repository

Below are details for each of the Jupyter notebooks in order of clarity and importance.

File	Description
shuffle-csvs	Main file responsible for pre-processing the data into managable chunks. Includes visualizations for data breakdown
mobilenet_master	Most MobileNet benchmarks performed using this file
imagenet_model_benchmarks	Varoius models trained and benchmarked on the QuickDraw dataset
cnn-rnn	MobileNet + LSTM model
wavenet	Implements the WaveNet architecture
mobilenet50epochs	Test run for Dropout vs recognized
autoencoder	Run autoencoders and variational autoencoders
ensemble	Construct ensemble of trained models

Training Model visualizations

Visualizations for training across models and epochs are present in the models directory.