Software Engineering project

API development using a Machine Learning model

Project requirements:

• Develop an API making use of a trained machine learning model.
• Use GIT version control concept through the API development.
• The API should include testing, code formating, logging and user login features.
• The API should be executable.
• Implement a CI/CD pipeline for additional credits.
• Documentation and presentation should be available upon submission.

API main goal:

• The API is to process a single digit (from 0 to 9) audio signal and return the corresponding predicted digit using ML model in the backend.
• The selected machine learning model for this project is the audio MNIST (Dataset, code) which identifies digits from audio inputs.

Disclaimer: the ML model was trained to a 94% test accuracy but does not generalize on all real life test cases due to the reduced dataset size. Training the ML model on additional / augmented data is out of the project'scope.

• The basic input method is through file selection. Additional developments are listed below if time permits..

API optional future developments (out of project'scope):

• Capture single digit audio signal from microphone.
• Capture multiple digit audio signal from microphone and return sequence of predicted digits.
• Use multiple digit audio prediction for user login.
• Use augmented / additional data to improve generalization on model prediction (male/female voices, accents, etc).

Getting started:

• Environment setup :

# Use virtualenv package to create a virtual python environment:
sudo apt-get install python-virtualenv
# Clone the repository:
git clone git@github.com:olivier-2018/SoftwareEngg_project.git  --branch development
# Create a virtual environment within the repo:
virtualenv venv
# Activate the virtual environment:
source venv/bin/activate
# Install dependencies:
pip install -r requirements.txt

• Set Flask environment variables:

Linux:
export FLASK_APP=run.py
export FLASK_ENV=development (or production, or testing as required)
export SECRETE_KEY="<whatever you want>" (optional, Flask will assign a secret hash if unset)

Windows powershell:
$env:FLASK_APP = 'run.py'
$env:FLASK_ENV = 'development' (or 'production', or 'testing' as required)
$env: SECRETE_KEY = <whatever you want> (optional, Flask will assign a secret hash if unset)

• Launch the API locally:

flask run

Note: on WSL you may need to export the display with an Xserver to run flask

• The API will automatically deploy to Heroku upon succesful build on the main branch.

Note: The Heroku app address is kept private not to reach the free account usage limit during the app development.

Testing:

• Unit and functional testing functions are located in the "tests" folder.
• Testing is automatic as part of the CI/CD pipeline but can also be launched manually using the command:

pytest -vrxXs

Demo

Illustrations:

Welcome screen

Sign-up screen

Welcome screen

Sign-in screen

App selection

API call and results

User profile screen

App version screen

Information screen

Tests

CI/CD & deployment