prabinrath / carlalidardatageneration

Repo for running carla sim, and generating training data for Lidar perception models.

CarlaLidarDataGeneration

Repo for running carla sim, and generating training data for Lidar perception models.

CarlaScenarioRunner()

Main class file for generating data from Carla Simulator
SEE RunCarlaScenarioRunner.py for example

Initialize with Constructor:

newClass = CarlaScenarioRunner(data_path = "file/path", intersection = "Intersection1")
data_path: path to data output folder
Valid Intersections: "Intersection1","Intersection2","Intersection3","Intersection4"

Spawn Sensors and Entities with:

spawn_lidar(model)
spawn_camera()
spawn_vehicles(num_vehicles)
spawn_pedestrian(num_pedestrians)

Run simulation with:

gather_simulation_data(sim_seconds)

ScenarioAnnotationVisualizer()

Visualization Class to display Lidar Pointclouds with drawn bounding boxes
SEE End of file for run example

drawBoundingPolygons.py

Additional code which takes in instance segmentized Camera images, draws bounding polygons around vehicles and pedestrians, saves all annotated images as single video file

Synchronous_mode.ipynb

Jupter Notebook used during code development, contains useful methods for testing the framework with Carla