DepthAI USB3 | Modular Cameras



Use DepthAI on your existing host. Since the AI/vision processing is done on the Myriad X, a typical desktop could handle tens of DepthAIs plugged in (the effective limit is how many USB ports the host can handle).


Board Layout

USB Layout

A. 5 V INE. Left Camera Port
B. USB3CF. DepthAI Module
C. Right Camera PortG. Myriad X GPIO Access
D. Color Camera Port

What’s in the box?

  • DepthAI USB3 Modular Cameras Carrier Board
  • USB3C cable (6 ft.)
  • Power Supply


Follow the steps below to setup your DepthAI device.

Connect your modular cameras.

The FFC (flexible flat cable) Connectors on the BW1098FFC require care when handling. Once inserted and latched, the connectors are robust, but they are easily susceptible to damage during the de-latching process when handling the connectors, particularly if to much force is applied during this process.

The video below shows a technique without any tool use to safely latch and delatch these connectors.

Connecting the Modular Cameras to BW1098FFC

Once the flexible flat cables are securely latched, you should see something like this:

BW1098FFC Connected to Modular Cameras

BW1098FFC modular camera top side

Connect your host to the DepthAI USB carrier board.

Connect the DepthAI USB power supply (included).

Install the Python DepthAI API.

See our instructions.

Calibrate Stereo Cameras.

Have the stereo camera pair? Use the DepthAI calibration script.

Download and run DepthAI Python examples.

We’ll execute a DepthAI example Python script to ensure your setup is configured correctly. Follow these steps to test DepthAI:

  1. Start a terminal session.
  2. Access your local copy of depthai.
     cd [depthai repo]
  3. Run python3
    The script launches a window, starts the cameras, and displays a video stream annotated with object localization metadata:

    object localization demo

    In the screenshot above, DepthAI identified a tv monitor (1.286 m from the camera) and a chair (3.711 m from the camera). See the list of object labels in our pre-trained OpenVINO model tutorial.