Open yangyi02 opened 1 month ago
An easy way is to use the inverse of an annotated static object. Since our object annotation is in the camera frame, so the camera pose is just the inverse of it.
I thought your object pose annotation is the transformation of current frame object pose w.r.t. the CAD coordinate coordinate framework. And different objects may have different canonical coordinate framework, hence inverse of different static objects may give different poses.
Am I interpreting wrong here?
It is not, actually the object pose is the transformation of current object w.r.t. the current camera frames. In other words, it is in camera frame rather than the world frame.
Hi, thanks for releasing this dataset.
May I ask if the camera pose is also released? If so, where can I find them?
Thanks.