ChiWeiHsiao/SphereNet-pytorch

Is kernel pattern in code correct?

Opened this issue · 1 comments

def get_xy(delta_phi, delta_theta):
return np.array([
[
(-tan(delta_theta), 1/cos(delta_theta)*tan(delta_phi)),
(0, tan(delta_phi)),
(tan(delta_theta), 1/cos(delta_theta)*tan(delta_phi)),
],
[
(-tan(delta_theta), 0),
(1, 1),
(tan(delta_theta), 0),
],
[
(-tan(delta_theta), -1/cos(delta_theta)*tan(delta_phi)),
(0, -tan(delta_phi)),
(tan(delta_theta), -1/cos(delta_theta)*tan(delta_phi)),
]
])

I have a simple question for your code.

In the sphereNet paper, they defined a kernel pattern as below

However, your code in line 20, the center is (1, 1) not (0, 0).

I'm confusing what is correct 😢

I noticed the same thing!