Hello everyone,
I am working on an HSE prevention project aimed at raising awareness about poor posture.
I am using Nuitrack with an Azure Kinect, and I would like to get some technical clarifications before moving forward.
My goal is to analyze, in real time, the angle of certain joints (back, neck, arms…) and to color the skeleton in green / orange / red depending on the posture risk level:
The objective is to make posture detection intuitive and visually clear for the user.
My questions
Q1. Which Nuitrack license is required for:
Q2. Is it possible to change the skeleton color directly through the Nuitrack API, or do I need to modify the provided examples (Nuitrack.NET / Unity)?
Q3. For detecting poor posture:
Q4. In your opinion, what is the best development approach:
-
modifying the SkeletonTracker example in C# (Nuitrack.NET),
-
or working directly in Unity with the Nuitrack SDK?
Q5. Are there any known limitations with Azure Kinect for this type of use (tracking, stability, joint accuracy)?
Thank you in advance for your help and for any documentation or code examples that could help me move forward.
Hello @Kalonek
- We offer two subscription plans to suit different needs:
-
AI Online Subscription - it gives you full access to the core software at the lowest price.
-
AI Creative Subscription includes everything in the Online plan (link), plus:
-
priority technical support (faster response times),
-
advanced integration with the most popular engines
-
early access to new samples
- Nuitrack does not provide direct API control for rendering (e.g., setting skeleton joint colors to green/orange/red).
The SDK only provides tracking data.
Visualization (including skeleton rendering and coloring) is done in:
- Unity,
- OpenGL / DirectX examples,
- or your own rendering pipeline.
So yes - you will need to modify the provided examples (e.g., Nuitrack.NET or Unity sample) or implement your own visualization logic.
- Nuitrack provides 3D joint positions and orientations, but it does not directly output ergonomic angles such as:
- neck flexion angle,
- torso inclination,
- shoulder/arm angles (in ergonomic terms).
Therefore:
You will need to compute angles yourself using joint coordinates (vector math between joints).
This is the standard approach for posture analysis pipelines.
- Both approaches are valid, but:
Unity + Nuitrack SDK is usually the best option if you want:
- fast prototyping,
- real-time visualization,
- easy UI/UX development,
- flexible rendering (color-coded skeleton, overlays).
C# Nuitrack.NET SkeletonTracker example is better if:
- you want a lightweight desktop application,
- or low-level control without a game engine.
For your use case (real-time posture visualization with color feedback), Unity is generally the more practical choice.
- Using Azure Kinect typical considerations are:
- Works well in controlled indoor environments.
- Accuracy decreases with:
- occlusions (arms crossing body, sitting positions),
- extreme poses,
- multiple people close together.