Intel Drone RTF (internal camera: Intel RealSense R200) +
Movidius Neural Compute Stick V1
Ubuntu 16.04 aero version
OpenCV v. 3.3.0
Error output running NCAPPZOO applications with live camera:
VIDEOIO ERROR: V4L2: Pixel format of incoming image is unsupported by OpenCV
libva info: VA-API version 0.39.0
libva info: va_getDriverName() returns 0
libva info: Trying to open /usr/lib/x86_64-linux-gnu/dri/i965_drv_video.so
libva info: Found init function __vaDriverInit_0_39
libva info: va_openDriver() returns 0
OpenCV Error: Unspecified error (GStreamer: unable to start pipeline...
yes, I used the procedure you mentioned and, in a second step, I removed the NCSDK v1 and I replaced it with v2 with the following procedure:
git clone -b ncsdk2 http://github.com/Movidius/ncsdk
The examples worked fine, using v1 and work fine also with NCSDK v2.
Then, I have installed ncappzoo:
git clone -b ncsdk2 https://github.com/movidius/ncappzoo.git
The ncappzoo's apps work fine, for example, also the stream_infer you mentioned work opening the camera using the command:
"v4l2src device=/dev/video" + CAMERA_INDEX
The apps that use live images/camera with cv2.VideoCapture command do NOT open the camera with the error message: VIDEOIO ERROR: V4L2
For example, this error is generated when I run live-image-classifier or video_face_matcher.
As tentative, I installed the OpenCV from source executing the install- opencv-from_source.sh (from video_face_matcher app)
It do NOT solve the problem.
I installed the OpenCV from source executing the install- opencv-from_source.sh (from video_face_matcher), removing the previous OpenCV, installing the OpenCV using Makefile with "-D WITH_V4L = ON".
This do NOT solve the problem.
Executing "sudo apt install v4l-utils", it's already exist.
Considering I'm using Ubuntu 16.04 OS on the Intel Drone, do you confirm that cv2.VideoCapture() commands should work with Intel Drone RTF's R200 camera?
If yes, I'll try to reinstall NCSDK and OpenCV.