From 3D model to AI on the Edge

A nice entry-level, practical introduction to synthetic data generation for AI training purpose. The examples use Unity (+ Perception) and Three.js (+ Playwright). Babylon.js is mentioned at 55:16 by yours truly :slight_smile:

3 Likes

Man! thanks a ton!!

Great talk! I wonder could you use some HUD bracket in 3D scene to automatically label the object?
Another thing that would be cool to add animations and iterate through it so ie. the forklift moves the fork up and down.

It should be fairly straightforward to programmatically extract the tagging info, and I would assume that there is a way to upload pre-tagged images via the Custom Vision API, but I haven’t tried so I don’t know for sure.