camera_stream

C++ Camera Stream

Overview

This example showcases the smallest possible DepthAIv3 pipeline written in C++ that streams raw color frames to the DepthAI Visualizer. It creates a single dai::node::Camera, requests an 1280x800 NV12 output from CAM_A, registers that output on a dai::RemoteConnection, and keeps the pipeline alive until you quit from the Visualizer. Use it as a template when you need a C++ starting point for experimenting with camera streaming or registering your own topics.

Demo

The application renders directly inside the DepthAI Visualizer. After you start it (see the steps below), open http://localhost:8082 and look for the images stream tile; focusing the browser tab lets you send hotkeys (including q) back to the app.

Peripheral Mode

Installation

Prepare a development environment with:

• DepthAI C++ library (depthai-core) installed and discoverable.
• CMake 3.20+.
• A C++17 capable compiler

To install the DepthAI C++ library, follow the instructions here. In short:

git clone --recurse-submodules https://github.com/luxonis/depthai-core.git
cd depthai-core
mkdir build
mkdir -p /opt/Luxonis/depthai-core
cmake -S . -B build -DBUILD_SHARED_LIBS=ON # Optionally use a custom installation path here like -DCMAKE_INSTALL_PREFIX=/tmp/depthai-core
cmake --build build --parallel
sudo cmake --install build # Sudo is not needed if you used a custom installation path above

Build & Run

cd cpp/camera_stream
cmake -S . -B build # In case you used CMAKE_INSTALL_PREFIX above, add -DCMAKE_PREFIX_PATH=/tmp/depthai-core
cmake --build build
./build/main

Keep the terminal open—the process stays alive while the pipeline runs. Open http://localhost:8082 in your browser to view the live images stream, and press q to stop the application cleanly.

Standalone Mode (RVC4 only)

Running the example in standalone mode builds and deploys it as an OAK app so that it runs completely on the device.

1. Install oakctl by following the instructions here.

2. Connect to your device:

   oakctl connect <DEVICE_IP>

3. From the cpp/camera_stream directory, run the packaged app:

   oakctl app run .

oakctl uses the provided oakapp.toml to build the C++ project inside the Luxonis base container and deploy it to the device. Configuration tweaks such as changing the camera resolution or registering more topics should be done in src/main.cpp, then re-run oakctl app run ..