Powered by:

This repo contains all the modules required to launch a Heroku App that classifies images into the ten categories from CIFAR10 dataset, the model was trained with convolutional neural networks using keras sequential wrapper on tensorflow.

Demo site: https://cnn-cifar10-bts.herokuapp.com/

Model training notebook is not included. But h5 file with the trained model is.

If you want to use it as the base for your own project you can do it by following the below instructions.

Requirements:

• Git
• Heroku
• Python 3.8.X or higher with pip3 installed
• See requirements.in for dependencies

Clone the repository into any directory you want by doing:

git clone https://github.com/techno1731/Keras_CNN_Cifar10.git

then cd into the Keras_CNN_Cifar10 folder and remove all git files. Linux example below:

In Linux or any UNIX like bash shell:

rm -rf .git*

Change into your virtual environment or create one for this project with your favorite tool, pyenv, venv, virtualenv, etc.

Use the package manager pip to install pip-tools.

pip install pip-tools

Then install the requirements into your virtual environment (recommended) by doing:

pip-sync

That's it you have an environment ready to develop with the codebase.

A demo of the application is available online at:

https://cnn-cifar10-bts.herokuapp.com/

To run the application locally after following installation instructions do:

streamlit run streamlit_deployment.py

The simply upload a picture and the app will return some strings with the result.

MIT