AI in Web: TensorFlow js
summary
Uses TensorFlow.js and models from TensorFlow Hub (MobileNet SSD v2, BlazePose 3D) to run browser-side inference at roughly 34 FPS, with model sizes from ~2–20 MB. Demonstrates a reusable ai.lab component that glues models to webcam/video/image sources, turning object detections into counters, zone-crossing alerts and pose-based interactions. Built demos include a sentiment aggregator that sends only labels (not images) to the server and a reliable people counter that visually marks bounding boxes.
AI-written summary · about
chapters (8)
- 0:00Start
- 4:17Machine Learning Applications
- 7:03JavaScript?
- 7:28WebGL Means JS + GPU
- 8:21Machine Learning Practicality?
- 15:16Solving Problems Over and Over
- 33:48Client Demo
- 36:01What's Next?
related talks
26mMachine Learning and AI for Angular Developers
34mAI and JavaScript, no You're Not Dreaming
41mTensorFlow.js 101: ML in the Browser and Beyond
30mBuilding Machine Learning Models on Edge with React
29mImage Processing in the Browser
20mTensorFlow.js Bringing Machine Learning to the Web and Beyond
46mArtificial Intelligence with JavaScript
54mDeep Learning in JS