Closed skmhrk1209 closed 1 year ago
Thank you for your question. For both static and dynamic objects, start_frame
and end_frame
denote how 3D bounding boxes are annotated, instead of the first and last frames where the instance is projected. More specifically, we split the dataset into windows of 200 meters for annotation, with each window distinguished by its start_frame
and end_frame
. A 3D bounding box is allocated in a window does not mean that it is visible in the full window.
Thank you for your reply. I understood that each instance is not necessarily visible in the window defined by start_frame and end_frame. So, is it right that each dynamic object is projected to just one frame denoted by timestamp
, each static object can be projected to any frames in the sequence, and the frames, where each static object is visible, can be identified by checking whether the instance ID is included in the instance ID map of each frame?
Yes, you understand it correctly. Thank you for the further clarification.
Thank you very much!
whether the instance ID is included in the instance ID map of each frame
hey man, could you tell me how to abtain the ID map?
@AmazingRoad Check the official documentation.
Could you explain what
start_frame
andend_frame
in the annotation files mean?According to the above official description about 3D bounding boxes, for static objects,
start_frame
andend_frame
can be interpreted as the first and last frames where the instance is projected, respectively. For dynamic objects, it is expected thatstart_frame
,end_frame
, andtimestamp
is the same, but it seems different. How can I interpretstart_frame
andend_frame
for dynamic objects?