Corrected HD Map - annotation data will be used, but the tool will attempt to correct any misaligned lines based on road intensity maps. This method is based on image processing and results may vary depending on the quality of the intensity maps. Intensity Map - only road intensity maps will be used. HD Map - only annotation data will be used and all generated lines will follow lines from the high-definition (HD) map.
The Line Source dropdown describes which data should be used to generate lines that will be visible by segmentation sensor. There are two main options that must be set before any other, more detailed, settings are available.
#CLIPPING DETECTOR PROJECT SIMULATOR#
To open the lane-line detector window, open a Simulator project in Unity editor, then select Simulator/Detect Lane-lines on the menu bar. Comparison of segmentation image with and without lane-lines is shown below. If line visibility for segmentation camera is unnecessary or undesired, you can safely ignore this tool. You can find more details about this on lane-line sensor page. If Generate Line Sensor Data in main options is enabled, output will also include correction data for the lane-line sensor. These meshes are not visible for normal cameras, but will be used by the segmentation sensor to differentiate between road and lane-lines. The main output from this tool is a set of meshes matching lane-lines on the road. Refer to the main options section for more details. On the other hand, if environment and annotations are slightly misaligned, you might want to try using the optional feature that will attempt to detect lines on intensity maps and correct the annotations based on this. If your annotation data is precise and aligns with the environment well, it's suggested to use it as the only input for this tool. Depending on the desired result, it can use data from map annotations and/or road intensity maps. The lane-line detector tool can used to prepare lane-line data for the segmentation and lane-line sensors.
Viewing and subscribing to ground truth obstacles.Modular testing with the Apollo AD stack.
#CLIPPING DETECTOR PROJECT WINDOWS#
0 kommentar(er)
