How to Create an Machine Learning Model from scratch and use it in an iOS app

This is a tutorial on how to create your own Machine Learning model from scratch and how to use it in your iOS app.

ML Model Type

This tutorial will be a YOLO (You Only Look Once) type of machine learning model as speed is a huge requirement for mobile devices. YOLO is one of the single stage detectors that runs the classifier once on the input image and do all of the work in a single pass.

Tools and Tech Stacks

Machine Learning Tools

• IBM Cloud Annotation - for labeling our dataset
• Jupyter Notebook - run our Python codes as well as creating, training, and saving our machine learning model
• Libraries - TuriCreate, Pandas

iOS Development Tools

• Xcode - best IDE for iOS development
• Libraries - CoreML, UIKit

Initial Setup for ML

Note: Only follow these instructions once. For updating or reopening, refer to the instructions under How to Update Machine Learning Model

1. With this repo downloaded, open Terminal and change directory to the project's directory
2. In Terminal, enter $ cd ML
3. If needed, install Python's virtualenv to have an isolated Python environment. $ pip install virtualenv
   • For a cheat sheet for virtualenv, click here
4. In Terminal, enter $ python3 -m venv ML_Model_Env to create a Python virtual environment on the ML_Model_Env
5. In Terminal, enter $ cd ML_Model_Env to go inside that directory
6. In Terminal, enter $ source bin/activate to activate Python environment. It should look like below

7. Install these libraries and its dependencies by running:
   • (ML_Model_Env) ➜ ML_Model_Env $ pip3 install turicreate
   • (ML_Model_Env) ➜ ML_Model_Env $ pip3 install jupyter
8. In Terminal, enter (ML_Model_Env) ➜ ML_Model_Env $ python3 -m notebook to launch Jupyter Notebook in your browser and then open Model Creator.ipnyb
9. Once you are done, enter in Terminal (ML_Model_Env) ➜ ML_Model_Env $ deactivate to clean state and remove all installed packages

How to Update Machine Learning Model

1. With this repo downloaded, open Terminal and change directory to the project's directory
2. In Terminal, enter $ cd ML/ML_Model_Env to go to ML_Model_Env directory
3. In Terminal, enter $ source bin/activate to activate Python environment. It should look like below

4. In Terminal, enter (ML_Model_Env) ➜ ML_Model_Env $ jupyter notebook to launch Jupyter Notebook in your browser and then open Model Creator.ipnyb

External Links

• IBM Cloud Annotations Introduction
• IBM Cloud Annotations Documentation

For questions or opportunities, contact me at:

• LinkedIn
• Email: samuelfolled@gmail.com
• MakeSchool portfolio - Expected gradutation date August 2021