TensorFlow.js 101: ML in the Browser and Beyond
summary
TensorFlow.js runs full ML in-browser and on Node with concrete results: COCO‑SSD (90 classes) does real‑time webcam detection, FaceMesh is a 3 MB model with 468 landmarks at ~25 FPS via WebGL, and BodyPix segments 24 body regions. Teachable Machine retrains a classifier in ~30 seconds from ~40 webcam samples and exports model.json. MobileNet v2 inference times match Python (≈8 ms), and converting pre/post processing to Node can yield 2× speedups as Hugging Face reports.
AI-written summary · about
chapters (9)
- 0:00Introduction to TF.js
- 1:46Using pre-trained models
- 5:50Creative model applications
- 9:24Transfer learning
- 12:37Cloud AutoML integration
- 15:55API structure and backends
- 18:33Performance benefits
- 23:53Community and resources
- 26:17Q&A session
related talks
20mTensorFlow.js Bringing Machine Learning to the Web and Beyond
37mTensorFlow.js - Bringing ML and Linear Algebra to Node.js
29mSuperpowers for next gen web apps: Machine learning
48mTensorFlow.js
24mTensorFlow.js + Node
34mA Powerful Introduction to Machine Learning with Js
15mMachine learning for next gen web apps with TensorFlow.js
31mIntroduction to Machine Learning in Tensorflow.js