Hi team, I’m trying to get the ego (aria) camera poses for each frame extracted from the take-level trimmed videos. However, the number of extracted frames from the videos is always 2 more than the number of provided poses in /<data_root>/annotations/ego_pose/train/camera_pose<take-uid>.json. For example, there are 3328 frames extracted from each camera of take cmu_bike01_2, but in ed3ec638-8363-4e1d-9851-c7936cbfad8c.json there are only 3326 poses under camera_extrinsics.
What’s the right alignment between the provided poses and frames?