NaN error while using map_pointcloud_to_image

#1

While using the ’ map_pointcloud_to_image’ with render_intensity =True, for LiDAR, why there is an error value of NaN. How to eliminate this error.

#2

Hi, can you report the error message here?

#3

Error message is

#  "541/20480 [..............................] - ETA: 1:21:11 - loss: nan - regression_loss: nan - classification_loss: nan
                # .... RuntimeWarning: invalid value encountered in true_divide
                # intensities = (intensities - np.min(intensities)) / (np.max(intensities) - np.min(intensities))
#4

I believe (np.max(intensities) - np.min(intensities) is 0, hence causing the error. Can you tell me the parameters you used to call map_pointcloud_to_image() ?

#5

map_pointcloud_to_image(self.nusc, pointsensor_token=lidar_pointsensor_token, camera_token=camera_token,render_intensity=True)

#6

Please tell me the value of lidar_pointsensor_token and camera_token so I can reproduce and fix it.