Rendering pose detection keypoints over the camera in react native

TensorFlow General Discussion
,
1

I am working on a project and using TensorflowJs in react native using tfjs-react-native library, I am perfectly able to detect the pose detection model using Movenet, but how can I render these keypoints over the camera in react native,
I have used SVG library to render circles but updating the states taking time at every frame, do you have any suggestions what is the best possible optimized way to render these keypoints over the camera?

3

Hi @harshbhatt7585 ,

I apologize for the delay in my response. Re-rendering keypoints in every SVG element is a time-consuming task that can lead to performance issues. You can consider using the WebGL backend to render keypoints directly on the camera feed. I’ve found a WebGL library called threejs that might be helpful for faster keypoint rendering.

Let me know if it helps. Thank You!!