Hi i noticed that in box_detection_train.json and box_detection_val.json, the q only contains 2d pixel coordinates, and let the model the re-build the 3d coordinates.
I wonder that do you provide camera parameters or something like that for the model?
Hi i noticed that in
box_detection_train.jsonandbox_detection_val.json, theqonly contains2d pixel coordinates, and let the model the re-build the3d coordinates.I wonder that do you provide camera parameters or something like that for the model?