Superpowers for next gen web apps: Machine learning
summary
Shows how TensorFlow.js runs full ML in the browser: use COCO-SSD for 90-object detection, Face Mesh 468 landmarks (3 MB) for real‑time face tracking at ~22–35 FPS, and BodyPix for segmentation. Demonstrations include invisibility, laser-eye AR, WebRTC+WebXR teleportation, and a clothing-size estimator that works in ~15 seconds. Models are trainable in-browser via Teachable Machine or exported from Cloud AutoML to TensorFlow.js; everything runs locally for lower latency and preserved user privacy.
AI-written summary · about
chapters (21)
- 0:00Intro
- 1:21Artificial Intelligence
- 2:15Machine Learning
- 3:40Common ML Use Cases
- 4:50Deep Learning
- 5:50Summary
- 6:58Visualising features
- 8:03Bad Features & Attributes
- 8:44Higher dimensions
- 10:07Use ML anywhere JavaScript runs
- 10:46Run, Retrain, Write
- 12:13Object Recognition
- 13:25Face Mesh
- 15:07Body Segmentation
- 15:38Invisibility
- 16:18Lasers
- 16:45Teleportation
- 21:01Cloud Auto ML
- 25:145 client side super powers
- 27:26Recommended Reading
- 28:17What will you make?
related talks
15mMachine learning for next gen web apps with TensorFlow.js
41mTensorFlow.js 101: ML in the Browser and Beyond
24mThe next level apps for Web developers Machine learning on browsers 🤯
19mML at Client Side
15mThe future of machine learning using JavaScript
41mPractical Machine Learning for Front-End Developers
34mA Powerful Introduction to Machine Learning with Js
20mTensorFlow.js Bringing Machine Learning to the Web and Beyond