What it does
This is a submission for the Picnic Hackathon. This project contains an image classifier using convolutional neural networks built using keras and tensorflow.
This model achieved 96% accuracy on training set.
README: Submission.tsv contains the submission file All other python files are helper and training files.
Enhance Dataset.py was used to enlarge the dataset. Move Files.py was used to manipulate the dataset structure into a format appropriate for Keras. Predict.py is used to predict the images in the test_images.tsv file
pre-trained-new.py were used to train the classifier. Run pre-trained-new.py to train the classifier.