This is a web application developed to detect and predict the handwritten digits written on a canvas and using tensorflow built model using softmax regression with Adam Optimizer. The model is quite good and is able to predict majority of digits.
Here is the deployed application: https://re-digit.herokuapp.com/
Steps to follow are:
1. git clone https://github.com/prakass1/digitrecognizer.git
2. cd --working_git_dir--
3. Ensuring all the requirements are satisfied as per requirements.txt, run python app.py
1. Make a fork of the application
2. git clone the app.
Example you could simply clone using git bash $git clone https://github.com/--username--/digitrecognizer.git
3. Make a new model, update it and build it using heroku.
Steps to build on heroku.
1. First make an account at heroku.
2. Download heroku cli for Windows/Linux/Mac
3. Using bash,
cd working_dir #containing the .git file
heroku create app_name
git remote -v
git push heroku master # Pushes the app to heroku master
Two important files that are necessary for herokus are:
a. Procfile # says what to deploy. Can leave it as is
b. requirements.txt # If any new requirements needs to be added it goes here