alexmach77 / number_recognition_tflow

Recognition of digits using the MNIST database (large database of handwritten digits)

##Steps

1. Create conda environment:

    conda create -n condaeup_flask_ml python=3.6.1
    source activate condaeup_flask_ml
   

2. Create requiremerents.txt file

    numpy
    pandas
    firefly-python
    Flask
    Flask-WTF
    tensorflow
    cv2
   

3. Install required packages:

    #cd number_recognition_tflow
    conda install --yes --file requirements.txt
    
    or
   
    #iterate over all lines in the requirements.txt file.
    while read requirement; do conda install --yes $requirement; done < requirements.txt
   

   If a requirement is missing, then use:

    pip install firefly-python
    pip install opencv-python
   

   If conda fails, we can also use pip:

    pip install -r requirements.txt
   

4. Running Flask

To run the application you can either use the flask command or python’s -m switch with Flask. Before you can do that you need to tell your terminal the application to work with by exporting the FLASK_APP environment variable:

a. For Mac:

$ cd webapp
$ export FLASK_APP=webapp.py
$ export FLASK_DEBUG = 1
$ flask run
 * Running on http://127.0.0.1:5000/

b. For Win:

$ cd webapp
$ set FLASK_APP=webapp.py
$ set FLASK_DEBUG = 1
$ flask run