Live video capture using Decklink devices. Video frames are processed with OpenCV and the stream is then forwarded to localhost using ZeroMQ.

Project used for the live augmented reality of surgical procedures using the DaVinci surgical system endoscope as video source.

Obtain a copy including all dependencies with:

git clone --recurse-submodules https://github.com/aledelmo/LiveCapture

Please refer to the OpenCV official guide for dependencies (e.g. libgtk2.0-dev).

Download DesktopVideo here for the latest BlackMagic drivers. Once downloaded, extract DesktopVideo and install the deb packages with the following instructions:

sudo dpkg -i desktopvideo_*.deb desktopvideo-gui_*.deb mediaexpress_*.deb

Please refer to the libzmq, cppzmq and opencv GitHub pages for more detailed instructions.

cd libs/ZMQ/libzmq
cmake -Hlibzmq-4.3.4 -Bbuild -DWITH_PERF_TOOL=OFF \
      -DZMQ_BUILD_TESTS=OFF -DCMAKE_BUILD_TYPE=Release \
      -DENABLE_DRAFTS=OFF .
cd build
make -j20

cd libs/ZMQ/cppzmq
cmake -H. -Bbuild -DENABLE_DRAFTS=OFF -DCOVERAGE=OFF \
      -D CMAKE_PREFIX_PATH=../libzmq/build
cd build
make -j20

cd libs/opencv
mkdir -p build && cd build
cmake -DWITH_FFMPEG=0 ..
make -j20

Compile client-side streaming service

mkdir -p build && cd build
cmake .. 
make -j20

Prepare server-side Python environment

pip install -r requirements.txt

1. Video ReadBack: Acquire video input on device with mode and forwarding the stream to the port

   ./Client -d <device id> -m <mode id> -h <port> [OPTIONS]

   OPTIONS: -3 <stereoscopic 3D> -p <pixel_format>

2. Signal simulation Generate test video stream on device with mode

   ./GeneratePattern -d <device id> -m <mode id> [OPTIONS]

   OPTIONS: -3 <stereoscopic 3D> -p <pixel_format> -c -s

3. Server: Request/Reply Python server receiving video stream

   python Server.py --port1 <port_number> --port2 <port_number>

Ubuntu 20.04 - Linux support only.

Tested using a DeckLink Duo 2 video acquisition card.

For any inquiries please contact: Alessandro Delmonte @ alessandro.delmonte@institutimagine.org

This project is licensed under the Apache License 2.0 - see the LICENSE file for details