orient = orient[argmax, :]
cos = orient[0]
sin = orient[1]
#img = Batch2Image(batch)
theta = np.arctan2(sin, cos) / np.pi * 180
theta = theta + centerAngle[argmax] / np.pi * 180
orientation_estimate = pydriver.common.functions.pyNormalizeAngle(np.radians(360 - info['ThetaRay'] - theta))
orientation_estimate = orientation_estimate / np.pi * 180
error = abs(orientation_estimate - info['Ry'])
if error > 180:
error = abs(360 - error)
error_lst.append(error)
Translation = Eval.GetTranslation(P, box_2D, orientation_estimate, dim)
distance_lst.append(np.linalg.norm(Translation - info['Location']))
#Translation = Eval.GetTranslation(P, box_2D, info['Ry'], np.array(dimGT))
#print box_2D, info['Ry'], dimGT
#print Translation
#print info['ID']
#print info['Location']
#print error
#print dimGT
#print dim
#sys.exit()
#error = abs(theta - info['LocalAngle'])
if i % 40 == 0:
print '===='
print info['Ry']
print info['ThetaRay']
#img = Batch2Image(batch)
theta = np.arctan2(sin, cos) / np.pi * 180
theta = theta + centerAngle[argmax] / np.pi * 180
orientation_estimate = pydriver.common.functions.pyNormalizeAngle(
np.radians(360 - info['ThetaRay'] - theta))
orientation_estimate = orientation_estimate / np.pi * 180
error = abs(orientation_estimate - info['Ry'])
if error > 180:
error = abs(360 - error)
error_lst.append(np.cos(np.radians(error)))
continue
Translation = Eval.GetTranslation(P,
box_2D,
orientation_estimate,
dim,
init=np.array(info['Location']))
#Translation = Eval.GetTranslation(P, box_2D, info['Ry'], np.array(dimGT))
distance_lst.append(np.linalg.norm(Translation - info['Location']))
#print box_2D, info['Ry'], dimGT
#print Translation
#print info['ID']
#print info['Location']
#print error
#print dimGT
#print dim
#sys.exit()
#error = abs(theta - info['LocalAngle'])
if i % 40 == 0:
print '===='
