You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@gkane26 can answer this better than I can, but quick answer since it's in my notifications -
init_inference prepares the model for use, creating the tensorflow objects, etc. depending on the type of model that's used. Some model types (tflite) need a frame in order to prepare the model, but it looks like it's just used to get a shape so it looks like that could be relaxed to also accept a shape tuple. Think the docs could spell that out a little clearer
if a frame is passed to init_inference, it calls get_pose, in which case it would return an estimated pose, otherwise it returns None.
get_pose is the main method for going frame -> pose.
it looks like there should be a check for initialization in get_pose so it raises a descriptive error if called before init, or else a little more permissive might be to auto-init with a warning saying "if u don't want to have first-frame lag call init first" or something. i'd be happy to PR that & docs if that sounds good.
Doubt
What's the diference between init_inference and get_pose?, according to the comments on code, I see that both return an estimated pose.
The text was updated successfully, but these errors were encountered: