Quick-start Guide to Gait Analysis with JavaScript and TensorFlow.js

Table of Contents
About the Author
About the Technical Reviewer
Acknowledgments
Preface
Chapter 1: Web Development
	Machine Learning Overview
	Web Communication
		Organizing the Web with HTML
	Web Development Using IDEs/Editors
		Building Blocks of Web Development
		HTML and CSS Programming
			Dynamic HTML
			Cascading Style Sheets
			Inline Style Sheets
			Embedded Style Sheets
			External Style Sheets
	JavaScript Basics
		Including the JavaScript
		Where to Insert JS Scripts
		JavaScript for an Event-Driven Process
	Document Object Model Manipulation
	Introduction to jQuery
	Summary
	References
Chapter 2: Browser-Based Data Processing
	JavaScript Libraries and API for ML on the Web
		W3C WebML CG (Community Group)
	Manipulating HTML Elements Using JS Libraries
	p5.js
		Drawing Graphical Objects
		Manipulating DOM Objects
		DOM onEvent(mousePressed) Handling
		Multiple DOM Objects onEvent Handling
		HTML Interactive Elements
			Interaction with HTML and CSS Elements
		Hierarchical (Parent-Child) Interaction of DOM Elements
		Accessing DOM Parent-Child Elements Using Variables
	Graphics and Interactive Processing in the Browser Using p5.js
		Interactive Graphics Application
		Object Instance, Storage of Multiple Values, and Loop Through Object
	Getting Started with Machine Learning in the Browser Using ml5.js and p5.js
	Design, Develop, and Execute Programs Locally
		Method 1: Using Python – HTTP Server
		Method 2: Using Visual Studio Code Editor with Node.js Live Server
	Summary
	References
Chapter 3: Human Pose Estimation in the Browser
	Human Pose at a Glance
		PoseNet vs. OpenPose
	Human Pose Estimation Using Neural Networks
		DeepPose: Human Pose Estimation via Deep Neural Networks
		Efficient Object Localization Using Convolutional Networks
		Convolutional Pose Machines
		Human Pose Estimation with Iterative Error Feedback
		Stacked Hourglass Networks for Human Pose Estimation
		Simple Baselines for Human Pose Estimation and Tracking
		Deep High-Resolution Representation Learning for Human Pose Estimation
	Using the ml5.js:posenet() Method
	Input, Output, and Data Structure of the PoseNet Model
		Input
		Output
		.on() Function
		Summary
	References
Chapter 4: Human Pose Classification
	Need for Human Pose Estimation in the Browser
	ML Classification Techniques in the Browser
	ML Using TensorFlow.js
		Changing Flat File Data into TensorFlow.js Format
	Artificial Neural Network Model in the Browser Using TensorFlow.js
		Trivial Neural Network
		Example 1: Neural Network Model in TensorFlow.js
		Example 2: A Simple ANN to Realize the “Not AND” (NAND) Boolean Operation
	Human Pose Classification Using PoseNet
	Setting Up a PoseNet Project
		Step 1: Including TensorFlow.js and PoseNet Libraries in the HTML Program (Main File)
		Step 2: Single-Person Pose Estimation Using a Browser Webcam
	PoseNet Model Confidence Values
	Summary
	References
Chapter 5: Gait Analysis
	Gait Measurement Techniques
	Gait Cycle Measurement Parameters and Terminology
	Web User Interface for Monitoring Gait Parameters
		index.html
	Real-Time Data Visualization of the Gait Parameters (Patterns) on the Browser
	Determining Gait Patterns Using Threshold Values
	Summary
	References
Chapter 6: Future Possibilities for Running AI Methods in a Browser
	Introduction
	Additional Machine Learning Applications with TensorFlow
		Face Recognition Using face-api.js
		Hand Pose Estimation
	Summary
	References
Conclusion
Index