Why Tensorflow object detection API uses YUV420SP

问题:

So I got the tensorflow object detection API running on Android and I've noticed while going through the code that before processing frames taken from the camera they're is a conversion that goes like this in the CameraActivity.java :

    imageConverter =
            new Runnable() {
                @Override
                public void run() {
                    ImageUtils.convertYUV420SPToARGB8888(bytes, previewWidth, previewHeight, rgbBytes);
                }
            };

I tried to look it up and I only understood the difference between the two types but I couldn't figure why this conversion is necessary (or preferable)..

Is this conversion since it's happening in real time and for every frame going to affect the preview or the processing time ?

Any information or explanation is much appreciated even if it's basic