How to modify code with fixed pose？

[YunnaChen]

Thanks for the great work. I would like to use real poses and fix poses . How should I proceed and is it possible to achieve? Any suggest will be very helpful, thank you!

[kg0221]

If I understand correctly what you're trying to do - use actual measured poses as inputs along with the video frames, there is a way to do that.

In motion_filter.track(), there are two calls to self.video.append(). The first passes in a pose of Id to initialize the pose to identity for the first frame. The next passes in a pose of None to run the algorithm without pose input information. You will need to pass your poses into motion_filter.track() and then pass that in the self.video.append() call instead of Id or None.

You'll need to make sure your pose has been properly converted to SE3.