To fix, add inference video path to dlc_config before saving in `reader.do_pose_estimation`
To fix, add inference video path to dlc_config before saving in
reader.do_pose_estimation