Apparatus and method for generating training data to train neural network determining information associated with road included in image
A training data generation method includes acquiring an image captured from a vehicle and location information of the vehicle corresponding to the image; acquiring map data corresponding to the acquired location information; determining truth data including information associated with a road included in the image from the acquired map data; and generating training data including the image and the determined truth data.
1. A training data generation method, comprising:
acquiring an image captured from a vehicle and location information of the vehicle corresponding to the image;
acquiring map data corresponding to the acquired location information;
determining truth data comprising information associated with a road included in the image from the acquired map data, wherein the truth data comprises a probability that the road is drivable, wherein the probability is determined based on an angle between the road and a direction corresponding to an orientation of the image, and wherein the orientation of the image corresponds to a front view from the vehicle; and
generating training data comprising the image and the determined truth data.
2. The training data generation method of claim 1 , wherein the acquiring of the map data comprises acquiring the map data by digital searching of a map database (DB) based on the acquired location information.
3. The training data generation method of claim 1 , wherein the acquiring of the map data comprises acquiring the map data based on the direction corresponding to the orientation of the image.
4. The training data generation method of claim 1 , wherein the truth data comprises a number of roads accessible by the vehicle when the vehicle moves forward.
5. The training data generation method of claim 1 , wherein the truth data comprises a gradient of the road determined based on first coordinates of a location of the vehicle corresponding to the image and second coordinates adjacent to the first coordinates.
6. The training data generation method of claim 1 , wherein the truth data comprises a probability that the road is matched to a predetermined road type.
7. A training data generation apparatus for generating training data from an image captured from a vehicle, the training data generation apparatus, comprising:
a processor configured:
to acquire the image and location information of the vehicle corresponding to the image;
to acquire map data corresponding to the acquired location information;
to determine truth data comprising information associated with a road included in the image from the acquired map data, wherein the truth data comprises a probability that the road is drivable, wherein the probability is determined based on an angle between the road and a direction corresponding to an orientation of the image, and wherein the orientation of the image corresponds to a front view from the vehicle; and
to generate training data comprising the image and the determined truth data.
8. The training data generation apparatus of claim 7 , wherein the processor is further configured to acquire the map data by digital searching of a map database (DB) based on the acquired location information.
9. The training data generation apparatus of claim 7 , wherein the processor is further configured to acquire the map data based on the direction corresponding to the orientation of the image and configured to determine truth data comprising a probability that the vehicle moves in the direction.
10. The training data generation apparatus of claim 7 , wherein the truth data comprises a probability that the road is matched to a predetermined road type.