Support for CPU/WebGPU backends in tfjs Handpose

[Zain Raza]

Hello everyone,

I am new to Tensorflow.js and I had a question about the Handpose model on the tensorflow/tfjs-models GitHub repository.

My understanding is that is currently only supports a WebGL or WebAssembly backend. However, do you think it might make a good contribution on GitHub to add support for a CPU or WebGPU backend as well?

The reason I ask is I have been using the model in a front-end web project recently, and when I used it the browser really started to slow down. It gave the following warning in the Inspect tool:

"tf-backend-webgl.js:6593 Your application contains ops that are small enough to be executed on the CPU backend, however the CPU backend cannot be found. Consider importing the CPU backend (@tensorflow/tfjs-backend-cpu) for better performance."

Would love to hear any advice folks have for implementing this feature or overcoming performance issues in general!