Multi-View Operating Room (MVOR) dataset consists of synchronized multi-view frames recorded by three RGB-D cameras in a hybrid OR during real clinical interventions. We provide camera calibration parameters, color and depth frames, human bounding boxes, and 2D/3D pose annotations. The MVOR was released in the MICCAI-LABELS 2018 workshop.
The value of ref_camera in annotations3D doesn't seem to be used anywhere, but the value itself is different, sometimes 0 sometimes 1 or 2. My question is does this value have any meaning? If not, in which coordinate system are the 3d keypoint annotations?
I would like to know why you choose to use only one keypoint for the head while coco usually use 5 (nose , 2x eyes , 2x ears). Was it because the persons in the database were wearing mask ?
Also what are you exactly pointing with this keypoint?
I'm trying to get de distance of a pixel at the depth image.
I guess the value of each pixel represents the disparity, and then, using the focal length (on annotation) and the baseline of the cameras (wich I dont know) we can calculate distance.
z = f*B/d
The questions are: my assumptions are right? if so what is the baseline of each camera?