Currently we rely on the semi robust OpenCV calibration. We need to implement Backpropagation to also use depth information and to better align the pointclouds.
We should at first use Ceres for CPU implementation and then maybe implement a GPU version if there is time.
Currently we rely on the semi robust OpenCV calibration. We need to implement Backpropagation to also use depth information and to better align the pointclouds. We should at first use Ceres for CPU implementation and then maybe implement a GPU version if there is time.