Occluded Objects about waymo-open-dataset HOT 5 CLOSED

Hi,
You can tell that by checking number of points inside a box. We provide tensorflow helper functions to compute that.
https://github.com/waymo-research/waymo-open-dataset/blob/master/waymo_open_dataset/utils/box_utils.py#L69

This field:
optional DifficultyLevel detection_difficulty_level = 5;
is supposed to be used to pass the difficulty level of a ground truth to evaluation. It is populated as LEVEL_2 if labelers believe that it is hard to label the box.

from waymo-open-dataset.

Hi @peisun1115 , I don't quite know how to tell if an object is occluded given this helper function
Suppose I only have a 3D bbox of a car that is occluded by a tree entirely. How can I tell if this car is blocked by calculating the points in that box?
Would appreciate it if you can give me a hint

from waymo-open-dataset.

I still do not understand how to do that?
The function has two arguments: box and points. From where can I get the points?
And how to decide later based on the returned number if the object is occluded or not?

from waymo-open-dataset.

You can use all points from that frame.

from waymo-open-dataset.

Hello,
I am interested only in the 2D visual data and wanted to use the "optional DifficultyLevel detection_difficulty_level = 5" field to deduce occlusion but it is marked "UNKNOWN" in all 5 segments I have tested. Any idea where can I find samples with this field?

from waymo-open-dataset.