Landmark detection using curve fitting for autonomous driving applications
In various examples, one or more deep neural networks (DNNs) are executed to regress on control points of a curve, and the control points may be used to perform a curve fitting operation—e.g., Bezier curve fitting—to identify landmark locations and geometries in an environment. The outputs of the DNN(s) may thus indicate the two-dimensional (2D) image-space and/or three-dimensional (3D) world-space control point locations, and post-processing techniques—such as clustering and temporal smoothing—may be executed to determine landmark locations and poses with precision and in real-time. As a result, reconstructed curves corresponding to the landmarks—e.g., lane line, road boundary line, crosswalk, pole, text, etc.—may be used by a vehicle to perform one or more operations for navigating an environment.
1. A method comprising:
determining, using one or more neural networks and based at least on sensor data obtained using one or more sensors of a machine:
one or more locations of one or more lines that represent one or more contours along one or more objects or one or more features; and
classification information corresponding to the one or more lines; and
performing one or more operations by the machine based at least on the one or more lines and the classification information.
2. The method of claim 1 , further comprising:
associating, based at least on the classification information, one or more class types with the one or more lines,
wherein the performing the one or more operations by the machine is based at least on the one or more lines and the one or more class types associated with the one or more lines.
3. The method of claim 1 , further comprising:
determining, based at least on the classification information, a first score associated with a first class type for a line of the one or more lines and a second score associated with a second class type for the line;
determining, based at least on the first score being greater than the second score, that the line is associated with the first class type,
wherein the performing the one or more operations by the machine is based at least on the line and the class type associated with the line.
4. The method of claim 1 , wherein the classification information indicates that the one or more lines represent at least one of one or more road markings, one or more lane lines, one or more road boundary lines, one or more intersection lines, one or more pedestrian walkways, one or more bike lane lines, text, one or more poles, one or more trees, one or more light posts, or one or more signs.
5. The method of claim 1 , wherein the one or more lines do not correspond to one or more bounding shapes of the one or more objects or the one or more features.
6. The method of claim 1 , further comprising:
determining, using one or more clustering techniques and based at least on the one or more lines, a final set of lines,
wherein the performing the one or more operations by the machine is based at least on the final set of lines and the classification information.
7. The method of claim 1 , wherein:
the one or more lines comprise at least a first line and a second line identified in an image;
the method further comprises:
determining that the first line is within a threshold distance to the second line; and
selecting the first line based at least on the first line being within the threshold distance to the second line; and
the performing the one or more operations by the machine is based at least on the first line and the classification information.
8. The method of claim 7 , wherein the selecting the first line comprises:
determining a first confidence associated with the first line;
determining a second confidence associated with the second line; and
selecting the first line based at least on the first confidence being greater than the second confidence.
9. A system comprising:
one or more processors to:
determine, using one or more neural networks and based at least on sensor data obtained using one or more sensors of a machine:
one or more curves corresponding to one or more contours of one or more objects or one or more features identified using the sensor data; and
classification information associated with the one or more curves; and
cause the machine to perform one or more operations based at least on the one or more curves and the classification information.
10. The system of claim 9 , wherein:
the one or more curves comprise at least a first curve and a second curve; and
the determination of the one or more curves comprises:
determining that the first curve is within a threshold similarity to the second curve; and
generating, based at least on the first curve being within the threshold similarity to the second curve, a third curve by combining the first curve and the second curve, the one or more curves including at least the third curve.
11. The system of claim 9 , wherein:
the one or more curves comprise at least a first curve and a second curve; and
the determination of the one or more curves comprises:
determining that the first curve is within a threshold distance to the second curve; and
generating, based at least on the first curve being within the threshold distance to the second curve, a third curve by combining the first curve and the second curve, the one or more curves including at least the third curve.
12. The system of claim 9 , wherein:
the one or more curves comprise at least a first curve and a second curve; and
the one or more processors are further to:
determine that the first curve is within a threshold distance to the second curve; and
select the first curve based at least on the first curve being within the threshold distance to the second curve,
wherein the machine is caused to perform the one or more operations based at least on the first curve.
13. The system of claim 12 , wherein the selection of the first curve comprises:
determining a first confidence associated with the first curve;
determining a second confidence associated with the second curve; and
selecting the first curve based at least on the first confidence being greater than the second confidence.
14. The system of claim 9 , wherein the one or more curves do not correspond to a bounding shape associated with the one or more objects or the one or more features.
15. The system of claim 9 , wherein the one or more processors are further to:
determine, based at least on the classification information, a first score associated with a first class type for a curve of the one or more curves and a second score associated with a second class type for the curve; and
determine, based at least on the first score being greater than the second score, that the curve is associated with the first class type,
wherein the machine is caused to perform the one or more operations further based at least on the first class type.
16. The system of claim 9 , wherein the determination of the one or more curves comprises:
determining, using the one or more neural networks and based at least on the sensor data, one or more locations of one or more control points identified using the sensor data; and
generating the one or more curves based at least on the one or more locations of the one or more control points.
17. The system of claim 9 , wherein the system comprises at least one of:
a system for performing simulation operations;
a system for performing simulation operations to test or validate autonomous machine applications;
a system for performing deep learning operations;
a system implemented using an edge device;
a system incorporating one or more Virtual Machines (VMs);
a system implemented at least partially in a data center; or
a system implemented at least partially using cloud computing resources.
18. One or more processors comprising:
processing circuitry to cause a machine to perform one or more operations based at least on one or more locations of one or more curves and classification information associated with the one or more curves, wherein the one or more locations of the one or more curves and the classification information are determined using one or more outputs of one or more machine learning models and based at least on sensor data obtained using one or more sensors of the machine,
wherein the classification information indicates that the one or more curves represent at least one of one or more road markings, one or more lane lines, one or more road boundary lines, one or more intersection lines, one or more pedestrian walkways, one or more bike lane lines, text, one or more poles, one or more trees, one or more light posts or one or more signs.
19. The one or more processors of claim 18 , wherein the one or more processors are further to:
determine, using one or more clustering techniques and based at least on the one or more curves, a final set of curves,
wherein the machine is caused to perform the one or more operations based at least on the final set of curves and the classification information.
20. The one or more processors of claim 18 , wherein the one or more processors are comprised in at least one of:
a processor for performing simulation operations;
a processor for performing simulation operations to test or validate autonomous machine applications;
a system for performing deep learning operations;
a system implemented using an edge device;
a system incorporating one or more Virtual Machines (VMs);
a system implemented at least partially in a data center; or
a system implemented at least partially using cloud computing resources.