Smart car operations are detailed including capturing a point cloud from a vehicle street view and converting the point cloud to a 3D model; applying a trained neural network to detect street signs, cross walks, obstacles, or bike lanes; and updating a high definition (HD) map with the neural network output.