RobotLocomotion / pytorch-dense-correspondence

Code for "Dense Object Nets: Learning Dense Visual Object Descriptors By and For Robotic Manipulation"

https://arxiv.org/pdf/1806.08756.pdf

Probably should load use dataset intrinsics in correspondence.

siyuanfeng-tri opened this issue 5 years ago · comments

siyuanfeng-tri commented 5 years ago

I think master is using the default K when computing correspondence.
https://github.com/RobotLocomotion/pytorch-dense-correspondence/blob/master/dense_correspondence/dataset/spartan_dataset_masked.py#L655
i added
K=self.get_camera_intrinsics(scene_name).K

also https://github.com/RobotLocomotion/pytorch-dense-correspondence/blob/master/dense_correspondence/dataset/dense_correspondence_dataset_masked.py#L117