Debugging your data

To set up the debugger, edit the path so that it points to your data in scale-example.py

if __name__ == '__main__':
    scene = create_scene(
            'data/', # <-- edit this path
            frames=range(1,6) # <-- edit frames if necessary
            )

Here are some things to check before you start debugging:

• Validate that the camera calibration is using all the important data (for example, if you’re using a fisheye camera, add the distortion coefficient and the camera model).

• If you are using a different point cloud format like .bin , you will need to change how the data is loaded.

• The lidar toolkit supports point clouds with 4 values per point (x, y, z, intensity), you may need to filter the point cloud if that is not the case.

• If the point cloud is in world coordinates, you will need to subtract the poses from it (the lidar toolkit will transform point clouds from ego to world coordinates using the poses).

If the data follows the data structure suggested and you finished the checks before debugging, you should be able to start debugging your data!

Validate your data

Scale’s lidar debugger also allows you to validate your data. To use the debugger, uncomment lines 8 and 9 shown below. You can comment out lines 12 and 14 to avoid creating a task when debugging.

if __name__ == '__main__':
    scene = create_scene(
        'data/',
        frames=range(1,6)
        )

    # Debugging methods
    scene.get_frame(index=1).add_debug_lines()    # add lines using the camera position and heading
    scene.preview()     # preview task

    # Upload data to S3 bucket
    # scene.s3_upload(S3_BUCKET, path='test-scale') # comment this line for testing
    # Create tasks/ Scale API request
    # scene.create_task(TEMPLATE).publish() # comment this line for testing

If everything looks correct after debugging, continue to the next step to create your task!