Object and Lane Detection

Lane boundary, pedestrian, vehicle, and other object detections using machine learning and deep learning

You can detect objects from vision data using machine learning and deep learning techniques. You can also segment, detect, and model parabolic or cubic lane boundaries by using the random sample consensus (RANSAC) algorithm. After you detect objects, use Automated Driving Toolbox™ functions to evaluate and visualize the detections.

You can also detect road lanes in lidar point clouds using a deep learning approach. Automated Driving Toolbox provides a lidar lane detection network trained on the K-Lane data set. The pretrained network enables you to detect a maximum of six lanes. You can also evaluate the performance of detector using different metrics, such as classification accuracy, precision, recall, and F1-score.

To detect lanes in lidar point clouds, download the Automated Driving Toolbox Model for Lidar Lane Detection support package from the Add-On Explorer. For more information on downloading add-ons, see Get and Manage Add-Ons.

Fore more information about processing lidar point cloud data and importing point clouds from Velodyne packet capture (PCAP) files, see Process Point Clouds.

Functions

expand all

Object Detection Using Vision Data with Machine Learning

`peopleDetectorACF`	Detect people using aggregate channel features
`vehicleDetectorACF`	Load vehicle detector using aggregate channel features
`acfObjectDetector`	Detect objects using aggregate channel features
`configureDetectorMonoCamera`	Configure object detector for using calibrated monocular camera
`acfObjectDetectorMonoCamera`	Detect objects in monocular camera using aggregate channel features
`trainACFObjectDetector`	Train ACF object detector
`objectDetectorTrainingData`	Create training data for an object detector
`vision.PeopleDetector`	(Removed) Detect upright people using HOG features
`vision.CascadeObjectDetector`	Detect objects using the Viola-Jones algorithm
`trainCascadeObjectDetector`	Train cascade object detector model

Object Detection Using Vision Data with Deep Learning

`vehicleDetectorFasterRCNN`	Detect vehicles using Faster R-CNN
`configureDetectorMonoCamera`	Configure object detector for using calibrated monocular camera
`fastRCNNObjectDetectorMonoCamera`	Detect objects in monocular camera using Fast R-CNN deep learning detector
`fasterRCNNObjectDetectorMonoCamera`	Detect objects in monocular camera using Faster R-CNN deep learning detector
`ssdObjectDetectorMonoCamera`	Detect objects in monocular camera using SSD deep learning detector
`yolov2ObjectDetectorMonoCamera`	Detect objects in monocular camera using YOLO v2 deep learning detector
`yolov3ObjectDetectorMonoCamera`	Detect objects in monocular camera using YOLO v3 deep learning detector (Since R2023a)
`yolov4ObjectDetectorMonoCamera`	Detect objects in monocular camera using YOLO v4 deep learning detector (Since R2022a)
`vehicleDetectorYOLOv2`	Detect vehicles using YOLO v2 Network
`trainYOLOv2ObjectDetector`	Train YOLO v2 object detector
`objectDetectorTrainingData`	Create training data for an object detector

Lane Boundary Detection and Modeling

`segmentLaneMarkerRidge`	Detect lanes in a grayscale intensity image
`findParabolicLaneBoundaries`	Find boundaries using parabolic model
`parabolicLaneBoundary`	Parabolic lane boundary model
`findCubicLaneBoundaries`	Find boundaries using cubic model
`cubicLaneBoundary`	Cubic lane boundary model
`computeBoundaryModel`	Obtain y-coordinates of lane boundaries given x-coordinates
`insertLaneBoundary`	Insert lane boundary into image
`fitPolynomialRANSAC`	Fit polynomial to points using RANSAC
`ransac`	Fit model to noisy data

Lane Detection Using Lidar Data with Deep Learning

`lidarLaneDetector`	Create lane detector for lidar data (Since R2023b)
`trainLidarLaneDetector`	Train lane detector for lidar data (Since R2023b)

Evaluate Detections

`evaluateObjectDetection`	Evaluate object detection data set against ground truth (Since R2023b)
`evaluateLaneBoundaries`	Evaluate lane boundary models against ground truth

Visualize Detections

`insertText`	Insert text in image or video
`insertShape`	Insert shapes in image or video
`insertMarker`	Insert markers in image or video
`insertLaneBoundary`	Insert lane boundary into image
`insertObjectAnnotation`	Annotate truecolor or grayscale image or video
`vision.DeployableVideoPlayer`	Display video
`vision.VideoPlayer`	Play video or display image

Topics

Get Started with Lidar Lane Detection Using Deep Learning
Use lidar lane detection network to detect road lanes.
Detect, Classify, and Track Vehicles Using Lidar (Lidar Toolbox)
Detect, classify, and track vehicles by using lidar point cloud data captured by a lidar sensor mounted on an ego vehicle.

Featured Examples

Ground Plane and Obstacle Detection Using Lidar

Detect the ground plane and find nearby obstacles in 3-D lidar data.

Open Script

Lane Detection in Lidar Data Using Deep Learning

Detect lanes in lidar point clouds using LLDN-GFC deep learning network.

Open Live Script

Visual Perception Using Monocular Camera

Construct a monocular camera sensor simulation capable of lane boundary and vehicle detections.

Open Live Script

Generate Code for Lane Marker Detector

Generate C++ code for lane marker detector and validate the functional equivalence using software-in-the-loop (SIL) simulation.

Open Script

Automate Testing for Lane Marker Detector

Automate the testing of a lane marker detector algorithm and generated code.

Open Script

Train a Deep Learning Vehicle Detector

Train a vision-based vehicle detector using deep learning.

Open Live Script

Generate Code for Vision Vehicle Detector

Generate deployable code for a monocular-camera-based vehicle detector and validate the functional equivalence with simulation.

Open Example

Automate Testing for Vision Vehicle Detector

Automate the testing of a vehicle detector and generated code.

Open Script

Track Multiple Vehicles Using a Camera

Detect and track multiple vehicles with a monocular camera mounted in a vehicle.

Open Live Script

Perception-Based Parking Spot Detection Using Unreal Engine Simulation

Build a bird's-eye-view map of a parking lot using semantically segmented images from the ego vehicle camera, and detect empty parking spots from the map.

Open Live Script

Perception Based Live Parking Spot Detection Using Unreal Engine Simulation

Develop a live parking spot detection system using deep learning and SLAM.

Open Live Script