Create ML for iOS Developers - A Beginner's Guide

Download a Core ML model for iOS from Apple’s Core ML models page, starting with mobile net v2, and learn that the same steps fit other image classification models.

Open the model in Xcode, test predictions and labels, and see how 224 by 224 input size and image orientation influence confidence during preview.

Implement a SwiftUI interface to cycle through images from the asset catalog, display the current image, and reveal predictions via a predict button for a mobile net model.

Resize a UI image to 224 by 224, convert it to a CV pixel buffer, and predict with Mobilenet v2 using Core ML.

Learn to display class label probabilities from an image classification model in iOS, including sorting by highest probability, using a dictionary of string to double, and formatting confidence as percentages.

Explore building an iOS app that lets you pick a photo from camera or library, display it, and predict whether it shows a dog or a cat using a button.

Download a small Kaggle cats and dogs dataset, understand train and test splits, use folders named cats and dogs as labels, and start training with Create ML.

Launch the Create app in Xcode, train a cats-versus-dogs image classifier on a two-class dataset, and evaluate it with testing accuracy before exporting the model for iOS integration.

Integrate a cats and dogs model into an iOS app, replace mobilenet, set 299x299 input, and test on a real device for reliable predictions.

Enable users to select a photo from the photo library with a photo picker. Load the image and run a Core ML model for prediction on a real device.

Launch a camera image picker in SwiftUI, manage optional UI image bindings, and test real-device photo captures for dog or cat images and model predictions.

Explore sentiment analysis and its emotional tone in text, distinguishing positive and negative sentiments with examples from products and financial news, and learn how to apply it to real data.

Identify and download a Kaggle sentiment analysis dataset for financial news headlines, then inspect a two column CSV with sentiment and text, and prepare column headings for training.

train a sentiment analysis model from a single CSV file with sentiment and text columns using Xcode playgrounds on macOS, applying ML text classifier in Create ML and exporting model.

Learn to extract data into positive, negative, and neutral folders using Python, preparing text datasets for Create ML training.

Learn to train a text classifier with Create ML using labeled training and test folders for neutral, negative, and positive classes, and compare Xcode playgrounds with a Python workflow.

Drag and drop the financial news sentiment model into a SwiftUI app, predict sentiment with Core ML, and color the view green, red, or gray by positive, negative, or neutral.

Learn how object detection recognizes and classifies objects in images and videos, and how it locates a traffic signal while labeling it as a traffic light.

Download a pre-trained object detection model from Apple's machine learning models page, choose between YOLOv3 and YOLOv3 tiny, and learn how to integrate it into your app.

Explore the starter project by running it to select a photo or capture one, note that the predict button does nothing, and anticipate learning object detection with Apple's vision framework.

Discover how the vision framework enables face and object detection, image classification, barcode and text detection, and image alignment in iOS apps, leveraging models and the vision module.

Perform object detection using the yolov3 tiny model with the vision framework, importing the core ml model, and extracting bounding boxes, labels, and confidence.

Display detected observations on screen by building an observation list view that shows each observation's label and confidence from the observations array, with bounding boxes.

Display bounding box overlays on detected objects using geometry reader and custom graphics, with overlays updating on image changes. Use YOLO v3 tiny and discuss training a custom model.

Build an object detection model to identify traffic lights in photos. Work with a single image or multiple images to detect multiple traffic signals in a scene.

Annotate images with visual ml to mark objects like traffic lights, export a Json file with coordinates for object detection, and adjust labels to ensure correct traffic light labeling.

Train a traffic light detector in Create ML using object detection with an annotation file and 11 images, then export the ML model for app integration.

integrate a traffic light object detector into an iOS app using Create ML, replacing the YOLO tiny model with the traffic light detector and validate detection under different lighting.

Explore a Kaggle car prices dataset and learn to filter for Toyota and Honda cars with under 60,000 miles and five years old, using the tabular api data frame calculations.

Explore representing and manipulating tabular data in iOS with data table and data frame APIs, reading csv or json, and preparing data for regression to predict car prices.

Learn to load a csv car dataset into a data frame, train an ML regressor with Create ML, and export a car price model for app use.

Filter the carvana csv with a data frame to train an ml model on cars under 60,000 miles and five years, Toyota or Honda.

Explore building an iOS app that filters cars by make and model (Toyota, Honda), selects year and mileage, and predicts price. See how miles affect price with Toyota Avalon 2018.

Extract the car name column from the data frame, remove duplicates, build a json array of names, and write it to a json file for the iOS app.

Create a Carvana demo app by loading a car json, displaying in a picker, and predicting prices with a Honda/Toyota model trained on last five years and under 60,000 miles.

Review the course coverage—from integrating models to training image-recognition and sentiment-analysis classifiers. Explore object detection with vision, tabular data predictions like car prices, and other classifiers to build smarter apps.