Kinect is dead. Here is the best alternative

It’s official: Azure Kinect is dead

In a recent and unexpected move, Microsoft announced the discontinuation of its popular Azure Kinect sensor, sending ripples throughout the tech and development community. Azure Kinect has been the backbone for many applications, particularly those requiring depth sensing and human skeletal tracking capabilities. With this announcement, numerous businesses and developers were left in the lurch, searching for alternatives to keep their applications functional.

For some developers, this may feel like déjà vu. After all, Microsoft stopped manufacturing the original Kinect for Windows in 2015 and the Kinect for Xbox One in 2017. Now, it’s the Azure Kinect’s turn to be discontinued in 2023. As Brian Heater of TechCrunch put it, “Microsoft can’t stop discontinuing Kinect.”

This time, Microsoft partnered with companies such as Orbbec and Analog, providing them with technology to develop newer depth cameras. However, the hardware options are limited, while the body tracking software has been deprecated and receives no further updates.

LightBuzz to the rescue

A few years ago, I wrote the go-to book on Kinect development. After Microsoft’s decision to axe the device, I decided to create something better: not just a replacement, but a tool that makes sense in the modern world. A tool that would bring body-tracking AI to every device, every camera, and every operating system. That’s how the LightBuzz SDK was born.

In this guide, we’ll explore the benefits of using this fast, accurate, cross-platform body tracking solution. I’ll also show you how to migrate your existing projects from Kinect to other depth cameras and mobile platforms.

A Wide Range of RGB & Depth Cameras

Kinect was great for its time, but it only worked with one sensor. Microsoft’s body-tracking software was tied to Kinect hardware, so developers had no flexibility.

LightBuzz takes a different approach. It can generate 3D skeleton data from any regular RGB camera, without the need for depth sensors. And if you decide to use depth cameras, you’re covered. The SDK supports over twenty camera models from multiple manufacturers — all through a single, unified API.

That means no more juggling SDKs or rewriting code for each device. Just plug in your camera and start tracking!

⏩ Smooth migration (with examples!)

With the LightBuzz SDK, the workflow is streamlined. It’s all about four key steps:

Here are some common examples of working with frame data.

1) Start a camera

To start a camera, all you need to do is specify the sensor type of your choice. It could be a webcam, a RealSense, etc.

Sensor sensor = Sensor.Create(SensorType.Webcam); // RealSense, LiDAR, etc.
sensor.Open();

2) Acquire frames

Capturing frame data is straightforward: subscribe to the FrameDataArrived event and access Skeleton, Color, Depth, and IMU information.

sensor.FrameDataArrived += (sender, frame) =>
{
    DateTime timestamp = frame.Timestamp;
    int width = frame.Width;
    int height = frame.Height;
    var colorData = frame.ColorData;
    var depth = frame.DepthData;
    var imu = frame.IMUData;
    var bodies = frame.BodyData;
};

3) Capture skeleton joint data

When it comes to accessing the body skeleton data, all you need to do is specify the joints of your choice and retrieve their 2D and 3D positions.

foreach (Body body in bodies)
{
    Joint neck = body.Joints[JointType.Neck];
    TrackingState = neck.TrackingState;
    
    Vector2 position2D = neck.Position2D; // Screen
    Vector3 position3D = neck.Position3D; // World
    Quaternion orientation = neck.Orientation;
}

When it comes to transforming from one coordinate system to another, the LightBuzz API is a breeze.

// Map from screen to world space:
Vector2 point2D = /* a screen-space point */;
Vector3 point3D = frame.Map3D(point2D); // World
// Map from world to screen space:
Vector3 point3D = /* a world-space point */;
Vector2 point2D = frame.Map2D(point3D); // Screen

Conclusion

See how simple it is? For those looking to streamline their motion-sensing applications and projects, transitioning from the Azure Kinect SDK to the LightBuzz SDK provides a clearer, more straightforward, and efficient programming experience. Embrace the ease and intuitiveness of LightBuzz for your next project.