The goal is to upload an image file from browser and predict whether it's a cat or a dog.
This repository contains a pre-trained classifier (classifier.h5), that was trained with a simple CNN. The training follows the tutorial in https://medium.com/nybles/create-your-first-image-recognition-classifier-using-cnn-keras-and-tensorflow-backend-6eaab98d14dd
The data that train.py expects in directory dataset can be found here: https://www.superdatascience.com/pages/machine-learning (Part 8, Section 40, CNN)
- script for training the classifier
- script for prediction using the trained classifier
- webserver GET
- webserver POST: upload image file to webserver
- allow saving the image or direct prediction
- handle duplicate filenames and non-existing target directory when saving uploaded image
- do actual prediction for that image file
- add anaconda environment description to the repository
- improve styling of index.html