The output of face detection network is standardized

I built a pipeline for face detection and recognition based on Python deepstream_test_2,The pipeline structure is as follows:

filesrc --> h264parse --> nvv4l2decoder --> nvstreammux --> nvinfer(pgie) --> nvinfer(sgie1) --> nvvideoconvert --> nvdsosd --> nvegltransform --> nveglglessink

I want to perform the following operations for each face image when the primary network outputs the results:

# scale pixel values
face_pixels = face_pixels.astype('float32')
mean, std = face_pixels.mean(), face_pixels.std()
face_pixels = (face_pixels - mean) / std

And then pass it to Facenet model.Therefore, I want to standardize the detected faces in pgie’s Sink Pad.

pgiesrcpad = pgie.get_static_pad("sink")
if not pgiesrcpad:
   sys.stderr.write(" Unable to get sink pad of pgiesrcpad\n")
pgiesrcpad.add_probe(Gst.PadProbeType.BUFFER, sgie1_sink_pad_buffer_probe, 0)

How can I implement it in method sgie1_sink_pad_buffer_probe?