Hello everyone,
We’re in the process of developing a web application that displays an MJPEG stream. Our goal is to identify specific objects in the video in real-time and perform pose estimation. Ideally, we’d like it to function in a way similar to fiduciary markers, but instead, leverage AR-bases positioning to enable markerless tracking.
Could anyone share insights on whether this is feasible using Three.js with AR plugin?
Which one👀 ?
Three.js itself is just a rendering library, just draws triangles - similarly to all other WebGL libraries. To do video processing you need a library dedicated to that (you can render the results / markers using three.js tho, that’s 100% feasible.)
It was last updated 6 years ago 
the chance of it working with current version of three.js (or if working at all, for that matter) is as high as the chance of @DolphinIQ finally agreeing to give react-three-fiber a try
There’s a chance of other AR helpers existing - but it’s good to always look at the last commit date / amount of issues on GH. Most OSS resources are created by enthusiasts and tend to die out over time (three.js contributors don’t maintain the linked AR lib, and three.js changed quite a lot in these past 6 years.)