Learn to create machine learning models on the browser
Computer Vision and Machine Learning have had significant advances over the past 5 years. However, much of it is shrouded behind complex and difficult to understand mathematics. TensorFlow.js achieves to solve this by abstracting away the math and providing users with a set of APIs that allow fast and efficient ML models to be embedded directly into the browser.
This talk will go over the basics of Computer Vision and Machine Learning. Participants do not need to have any prior experience with computer vision, machine learning, or TensorFlow.js to attend. Knowledge of basic web development are recommended, but not required.
How to Attend
More information about TensorFlow.js extensions can be found on the TensorFlow Website.
You can familiarize yourself with basic web development using these resources:
The lab will walk you through creating a simple website that uses a computer’s webcam to get a live video feed, uses Posenet (a pretrained TensorFlow.js model) to identify a pose, and then output the resulting pose to the console.
Task 1 - Setting up a website
- Create an
- Creat a script file (e.g.
main.js) and link that to your
- Import the appropriate TensorFlow.js scripts:
<script src="https://cdn.jsdelivr.net/npm/@tensorflow/[email protected]/dist/tf.min.js"></script>` <script src="https://cdn.jsdelivr.net/npm/@tensorflow-models/posenet"></script>
Task 2 - Add an Video to your Website
- On Safari, you’ll have to allow media capture on insecure sites:
Develop -> Web RTC -> Allow Media Capture on Insecure Sites
Task 3 - Run the Posenet Convolutional Neural Network to Identify the image
- Check out the reference code in the
- The Posenet GitHub repository has very good documentation.