OpenCV4Tegra Version with Gstreamer and Opencv support

Hi, from 2 months I am trying to “intercept” a Gstreamer pipeline with opencv.
On my regular laptop I am able to to that using something like that:

input = "v4l2src device=/dev/video0 ! video/x-raw, framerate=30/1, width=640, height=480, format=RGB ! videoconvert ! appsink"
output = "appsrc ! videoconvert ! x264enc pass=qual quantizer=20 tune=zerolatency ! rtph264pay ! udpsink host=127.0.0.1 port=65001 sync=false"

cap = cv2.VideoCapture(input)
out = cv2.VideoWriter(output, 0, 30, (640, 480), False)

ret, frame = cap.read()
out.write(frame)

In this way, I can elaborate the frames flowing on the pipeline and, at the same time, continue the streaming.
On my laptop I can do this since I compiled OpenCV with “WITH_V4L=ON” and “WITH_GSTREAMER=ON”.
On my Tx2, if I print(cv2.getBuildInformation())
the output says me: GStreamer: NO; V4L/V4L2: NO/YES

So, is there a way to have a OpenCV4Tegra compiled version with gstreamer interface?

Thanks in advance

Opencv4tegra is built without gstreamer support.
You may build your own opencv library enabling gstreamer-1.0 (only) support.
If your code is tied to opencv2, you may build your own 2.4.13, but performance may be less.

You can get good performance with opencv3. Current version is 3.3.1. (Note that the opencv-3.3.1 version included in R28.2 preview is built without gstreamer nor cuda support. You have to build your own). You may check this (it was for opencv-3.3.0, you may adjust): https://jkjung-avt.github.io/opencv3-on-tx2/.