Doubt in 3d box construction

[arshemii]

In the following method:

def _construct_box_3d(self, l, h, w):
"""
Compute the coordinates of the vertices of a 3D bounding box in an
object coordinate system.
"""
x_corners = [0.5l, l, l, l, l, 0, 0, 0, 0]
y_corners = [0.5h, 0, h, 0, h, 0, h, 0, h]
z_corners = [0.5*w, w, w, 0, 0, w, w, 0, 0]
x_corners += - np.float32(l) / 2
y_corners += - np.float32(h)
z_corners += - np.float32(w) / 2
corners_3d = np.array([x_corners, y_corners, z_corners])
return corners_3d

Why x is calculated from l
and why z is calculated from w?

Shouldn't be the opposite?

[Nicholasli1995]

In the following method:

def _construct_box_3d(self, l, h, w): """ Compute the coordinates of the vertices of a 3D bounding box in an object coordinate system. """ x_corners = [0.5_l, l, l, l, l, 0, 0, 0, 0] y_corners = [0.5_h, 0, h, 0, h, 0, h, 0, h] z_corners = [0.5*w, w, w, 0, 0, w, w, 0, 0] x_corners += - np.float32(l) / 2 y_corners += - np.float32(h) z_corners += - np.float32(w) / 2 corners_3d = np.array([x_corners, y_corners, z_corners]) return corners_3d

Why x is calculated from l and why z is calculated from w?

Shouldn't be the opposite?

Hi, thank you for your interest in this repo. The computation of the xyz coordinates actually defines how a 3D box at a canonical pose is placed with respect to the coordinate frame. Imagine the x-axis points along the "length" direction of the box, and thus the x coordinates of the corners are related to l.