InferenceJS: Real-time computer vision in your browser
summary
Roboflow shows running computer vision models inference in the browser using InferenceJS as a thin layer over TensorFlow.js and Microsoft COCO models from Universe. It demos a rapid fine‑tune flow (Roboflow Rapid) that builds a detector and returns a working model on video in about 30 seconds and promises end‑to‑end label→train→deploy pipelines in under five minutes. The presenter runs a webcam demo and a QR‑scannable scavenger hunt that largely works though occasional network lag appears.
AI-written summary · about
chapters (9)
- 0:00InferenceJS intro and real-time CV
- 1:48Computer vision overview
- 3:14What is Rooflow
- 5:01Rooflow products
- 8:31Introducing InferenceJS
- 9:55Live coding demo
- 11:18Rapid AI model building
- 13:01Scavenger hunt demo
- 14:07Next steps
related talks
41mAI in Web: TensorFlow js
41mTensorFlow.js 101: ML in the Browser and Beyond
34mA Powerful Introduction to Machine Learning with Js
30mBuilding Machine Learning Models on Edge with React
56mTensorFlow.js: Make a smart webcam in JS with a pre-trained ML model
29mSuperpowers for next gen web apps: Machine learning
49mReal Time Face Detection through a Webcam Using AI
48mTensorFlow.js