I want to run tensorFlow JS model in a framework that have frontend and backend what are you recommendations?

Question

Currently we have a web app (single html page with a script tag) which runs a TensorFlow JS model, that we trained using teachable machine website.

The code runs correctly, and we can see the predictions.

Now, and according to our knowledge transferring into a framework that have frontend and backend will be better, so:

1. We will transfer the logic there.

2. And to add a database to store the prediction.

Which solution you suggest us to procced with to achieve our target in the most suitable way, currently we are thinking about (react JS + node JS) but we still can't determine if it's the right option.

Answer 1

The React + Node.js stack you're looking for is implemented in Remix (not affiliated). You are able to define both backend and frontend logic in the same file which is then compiled for both the browser and Node:

export async function action({request}) {
  // Server side
  const form = await request.formData();
}

export default function Home() {
  // Client side
  return <p>Hello World</p>
}