How to train Object Detector including sub-labels?

Here I am trying to build object detector(YOLO v2) and I want to train the models that detect a vehicle and wheells that are parts of the car.

I used image-labeling tool of matlab and made a labeled data of vehicles and wheels(Wheels is a sublabel of vehicle), where I followed the discription about image labeling below.

https://jp.mathworks.com/help/vision/ug/get-started-with-the-image-labeler.html?lang=en

However, I made " gTruth " and put it through "objectDetectorTrainingData(gTruth) ", and execute the training like here ;

[detector,info] = trainYOLOv2ObjectDetector(cds,lgraph,options), this detector only detects a vehicle and cannot detect wheels.

How can I train not only the main label( such as car), but also the sub labeled object(such as Wheel) ?