3D Box orientation Regression problem

[weisili2016]

hi,
in your code , global object orientation is regressed by CNN , from only the contents of the 2D bboxs result,
but , I found that the author of the paper[https://arxiv.org/abs/1612.00496] suggested that ,local orientation is regressed, not the global object orientation.
Did I get it wrong?

[smallcorgi]

Hi @weisili2016 ,
The CNN will predict the local object orientation, not the global object orientation.

So in testing phase, θlocal will add with θray.

3D-Deepbox/main.py

Line 252 in c409d46

line[-1] = angle_offset +np.arctan(float(line[11]) / float(line[13]))

[weisili2016]

@smallcorgi
data_processing.py
line 14.
def compute_anchors(angle):
....

so, in training phase ,the input angle is the local object orientation?

[smallcorgi]

@weisili2016
Yes.