Main Content

evaluate

Evaluate performance of lidar lane detector

Since R2023b

collapse all in page

Syntax

metrics = evaluate(detector,detectionResults,groundTruthData)

Description

metrics = evaluate(detector,detectionResults,groundTruthData) evaluates the performance of the lidar lane detection network utilizing global feature correlation (LLDN-GFC) detector, detector, based on its classification accuracy, precision, recall, and F1-score metrics. Use these metrics to measure the lane detection results, detectionResults, against the ground truth data, groundTruthData.

Note

This functionality requires Deep Learning Toolbox™, Lidar Toolbox™, and the Automated Driving Toolbox™ Model for Lidar Lane Detection support package. You can download and install the Automated Driving Toolbox Model for Lidar Lane Detection from Add-On Explorer. For more information about installing add-ons, see Get and Manage Add-Ons.

Input Arguments

collapse all

LLDN-GFC lane detector, specified as a lidarLaneDetector object.

Detection results for a collection of point clouds, specified as a P-by-2 table with these columns. P is the number of point clouds.

  • laneDetections — Locations of the lanes detected within a point cloud, specified as an M-by-3 matrix. M is the number of detected lane points. Each row in the matrix represents the [x y z] coordinates of a lane point.

  • labels — Labels of the detected lane points within a point cloud, specified as a M-by-1 categorical array. M is the number of detected lane points.

Data Types: table

Labeled ground truth data of the point clouds, specified as a P-by-Q table. P is the number of point clouds, which must match the number of point clouds specified to the detectionResults table. Q is the number of lane classes. Each column of the table represents a single lane class, such as lane1, lane2, or lane3. The names of the columns must match the names of the classes stored in the ClassNames property of the detector.

Each cell of the table must be empty or an N-by-3 matrix that specifies the [x y z] coordinates of N lane points of the corresponding lane in a point cloud data. An empty cell indicates that the point cloud associated with that row does not have any lane points for the corresponding lane.

Data Types: table

Output Arguments

collapse all

Performance metrics of the LLDN-GFC lane detector, returned as a structure with these fields:

  • Accuracy — Classification accuracy of the lane detector, returned as a scalar in the range [0, 1].

  • ClsF1Score — F1-score for classification, returned as a scalar in the range [0, 1]. This metric evaluates correct classification of lane points.

  • ConfF1Score — F1-score for confidence, returned as a scalar in the range [0, 1]. This metric evaluates correct detection of lane points.

  • Precision — Precision value for detection of lane points, returned as a scalar in the range [0, 1].

  • Recall — Recall value for detection of lane points, returned as a scalar in the range [0, 1].

References

[1] Paek, Dong-Hee, Seung-Hyun Kong, and Kevin Tirta Wijaya. “K-Lane: Lidar Lane Dataset and Benchmark for Urban Roads and Highways.” In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 4449–58. New Orleans, LA, USA: IEEE, 2022. https://doi.org/10.1109/CVPRW56347.2022.00491.

Version History

Introduced in R2023b

See Also

| |

Topics