Increasing performance at Odroid C2

[qlkvg]

Sup, guys.
Cause libORBEC.so for aarch64 is still unavailable, I started to use pre-compiled openni2 for ARM architecture with Odrobian Jessie on my Odroid C2. And fps is confusing. I configured 640x480 depth stream. When camera directed to close objects(about 1-1.5 meters) fps is about 12-15, but when it directed to far object(e.g. wall in 5-6 meters above camera), fps dropping to 1-2 and sometimes stucks for several seconds. Also, when I place my hand right near camera(40-50 centimeters), fps increasing to 30.
I tried to increase performance in several ways. Cropping image (I realy need only center of image) does not work at all. Setting max depth value to 10000 work, but do not giving any increase.
So, is there way to increase performance with Odroid C2? My test code is belong.
Thanks.

import cv2
import numpy as np
from primesense import openni2
from primesense import _openni2 as c_api
import ctypes
openni2.initialize(‘/home/qlkvg/Downloads/OpenNI-Linux-x86-2.2/Redist’) # can also accept the path of the OpenNI redistribution

dev = openni2.Device.open_any()
print “\n”.join(dir(dev))

depth_stream = dev.create_depth_stream()
depth_stream.start()
depth_stream.set_video_mode(c_api.OniVideoMode(pixelFormat = c_api.OniPixelFormat.ONI_PIXEL_FORMAT_DEPTH_100_UM, resolutionX = 640, resolutionY = 480, fps = 30))

this does not work

depth_stream.set_cropping(100,100,300,300)

setting max_depth value

depth_stream.set_property(0x1080FF41, ctypes.c_int(10000))
while True:
frame = depth_stream.read_frame()
frame_data = frame.get_buffer_as_uint16()
img = np.frombuffer(frame_data, dtype=np.uint16)
img.shape = (1, 480, 640)
img = np.concatenate((img, img, img), axis=0)
img = np.swapaxes(img, 0, 2)
img = np.swapaxes(img, 0, 1)
cv2.imshow(“image”, img)
cv2.waitKey(34)
openni2.unload()