Why do we invert the matrices in the proj_to_velo function in input_velodyne.py

[Megha Maheshwari]

Why do we invert the matrices in the proj_to_velo function in input_velodyne.py

[smanshei]

Hi, In kitti raw dataset we have labels in left camera frame. so we need a transformation from camera frame to velodyne frame.

[Megha Maheshwari]

Thanks. Could you also clarify, which angle from the bin file are you trying to take in the function
def filter_camera_angle(places):
"""Filter camera angles for KiTTI Datasets"""
bool_in = np.logical_and((places[:, 1] < places[:, 0] - 0.27), (-places[:, 1] < places[:, 0] - 0.27))
# bool_in = np.logical_and((places[:, 1] < places[:, 0]), (-places[:, 1] < places[:, 0]))
return places[bool_in]
and why are you restricting to this field. Is it because you want to align it to the left camera angle?

[smanshei]

yes it is filtering out the points which are outside the camera view angle