GOOGLE MLKIT Pose Detection says here:
The base SDK runs in real time on modern phones like the Pixel 4 and iPhone X. It returns results at the rate of ~30 and ~45 fps respectively.
However, on Android Emulator the CameraController.startImageStream returns 15-16 fps which means that while detecting poses with Google Pose Detection or drawing key points on a person's body, I am not getting 30 fps from the camera to even work properly with it.
One approach I am thinking about it doing CameraController.takePicture 30 times a second and draw on body points & detect poses using the frames. Additionally, this approach can help us control how many frames we need to capture per second, which we do not control with CameraController.startImageStream because it may work too slow or too fast. Would this approach would affect performance badly?