Open OneManArmy93 opened 5 years ago
May be there are no objects in the ground truth. Try with the frame which has some cars in it
Yes im working with a frame where there is a car.
I managed to find the root of the problem, it is in because the z_logical
is alwayse False
hence xyz_logical
is always False
so sphere_center, train_corners
are always empty. I noticed the same behavior in center_to_sphere(places, size, resolution=0.50, min_value=np.array([0., -50., -4.5]), scale=4, x=(0, 90), y=(-50, 50), z=(-4.5, 5.5))
because places[:, 2] < z[1]
this is always False
Can you explain the reason of using x_logical, y_logical, z_logical
and why those values, because they are genetaring the issue. Thx
EDIT: I changed the z values to z=(-4.5, 20)
and now it is not returning empty variables, but does that change affects anything?
z is the vertical direction of point cloud, you can set it as z=(-3, 1) for kitti data set, actually it is filtering out the ground truth objects which are out side the range of experiment range.
ok, i'll try to do that but i set it at (-4.5, 20)
and returned not empty viarbles. I guess i'll see how it goes from there.
Regarding my question I post it on an other issue:
what does this function dor create_objectness_label
?
why there is 3 loss functions?
and I couldn't find the declation of the cordinate
function and What does it do?
Thank you
I tired to execute the this funtion
def create_labels(places, size, corners, resolution = 0.5, x=(0, 90), y=(-50, 50), z=(-4.5, 5.5), scale=4, min_value=np.array([0., -50., -4.5])):
but the returned variablessphere_center, train_corners
have the size of(0,3) and (0,8,3)
but they are empty. Any idea about that? Thank you