Some questions about the 3d detection in KITTI dataset

[wangx1996]

Hello, author:
Firstly ,thanks for this great work. When I read the code , I got some questions about the 3d detection part.
1.why the target which made for training doesn't use the trigonometric function to encode the angle offset?

          if alpha < np.pi / 6. or alpha > 5 * np.pi / 6.:
            rotbin[k, 0] = 1
            rotres[k, 0] = alpha - (-0.5 * np.pi)    
          if alpha > -np.pi / 6. or alpha < -5 * np.pi / 6.:
            rotbin[k, 1] = 1
            rotres[k, 1] = alpha - (0.5 * np.pi)

but the decode part still use actan2 to get the angle?

  idx = rot[:, 1] > rot[:, 5]
  alpha1 = np.arctan2(rot[:, 2], rot[:, 3]) + (-0.5 * np.pi)
  alpha2 = np.arctan2(rot[:, 6], rot[:, 7]) + ( 0.5 * np.pi)

2. what's the use of the 'gt_det'?