react-xr VS mind-ar-js

I think the easiest for you would be to get into this through web development try this https://github.com/pmndrs/react-three-fiber coupled with this https://github.com/pmndrs/react-xr But don't expect super beautiful graphic from this, it's abstraction over three.js library which is abstraction over webgl standard which implements OpenGL ES 2.0 which is 7 years old graphics standard targeted to mobile devices with not much power. Or you can experiment with unity (unity XR) or unreal engine which are harder to learn but produce better graphics.

there's also a VR layer that makes previously complex things so easy https://github.com/pmndrs/react-xr

I'm a pure threejs afficionado myself, but r3f also has some good things going for it: https://github.com/pmndrs/react-xr

as an example, these are AR/VR components that allow you to build apps in minutes because all the hard parts have been taken take of, for instance gestures: https://github.com/pmndrs/react-xr

I haven’t played with it, but if I was going to I’d start here https://github.com/pmndrs/react-xr

React-XR for leveraging the WebXR API (e.g VR Headset + controllers) & React-Three-Drei which has useful functions for creating WebGL content.

React three fiber is amazing for building 3d on the web, and has an XR canvas https://github.com/pmndrs/react-xr

Their docs and examples are beautiful and make it so easy to get started. As a web dev myself, the transition was pretty easy, since it was as simple as "swapping dom out for threejs objects". You get the react "engine" around it for free. Super fun for game development. (even with WebXR support https://github.com/pmndrs/react-xr)

And actually OP, if all you need is image-tracking you could look into JS-lib-only solutions like this one: https://github.com/hiukim/mind-ar-js

On WebAR MindAr with AFrame is your best choice , javascript again Check the code and demos https://github.com/hiukim/mind-ar-js

Mind AR. It's awesome and open source. Used it to build my own custom web-based React app on top of it (lively.so).

I'm working on a project and I need MindAR (link) as the main framework. I wrote some automation on HTML files we deliver to our customers, but as you may know, MindAR requires some sort of compilation on the target images (the compiler is here).

You could try mind-ar https://github.com/hiukim/mind-ar-js it's got face tracking built-in, is browser-based and you can build the 3d stuff with Aframe https://aframe.io/ it's pretty easy if you know HTML/javascript.