A Meta Quest application that uses computer vision and Poker Odds API to calculate poker hand strenght in real time.

• This project uses computer vision to scan poker game cards and runs simulation engine to calculate hand equity in real-time.
• It provides instant, data-driven insights direcly in your field of view, helping players make mathematically optimal decisions.

• Unity 6000.0.55f1
• Meta XR SDK
• Meta PCA Sample
• Roboflow Inference
• Poker-Odds NodeJs -> Serverfile under ./backend/server.js

git clone https://github.com/nigelhartm/MetaPCARoboflow.git

👉 https://github.com/roboflow/inference

To run AI inference, you need to start a local Roboflow Inference Server on your computer (PC or Mac). This project does not run the model directly on the Meta Quest or on-device. Instead, it sends images from Unity or your headset to the inference server running on your desktop. For CUDA I recommend this link https://developer.nvidia.com/cuda-downloads

• Initialize Node.js environment in the backend folder:

npm install

• Start the server:

node server.js

• The Node.js backend handles communication between Roboflow inference and your Unity/Meta Quest app.

• Create a file Assets/Secrets/APIKeys.cs with your API key:

public static class APIKeys
{
    public const string RF_API_KEY = "your-roboflow-api-key";
}

• Open Main.unity in Unity.
• Update the IP addresses in RoboflowCaller.cs to match your local inference servers:

[SerializeField] private string LOCAL_SERVER_IP_ADDRESS = "http://YOUR_COMPUTER_IP:9001";        // Main scene
[SerializeField] private string LOCAL_SERVER_POKER_IP_ADDRESS = "http://YOUR_COMPUTER_IP:3000"; // PokerOdd scene

• Build the project for Android (XR Plugin Management > Oculus).
• Deploy and run on Meta Quest with permissions for camera and local network access.

⚠️ Server not running: Don't forget that the server need to be started before and the first call takes up to a minute to download the model before!

• WebCamTextureManager - Use the Passthrough Camera of Meta Quest
• EnvironmentRaycastManager - Using Meta’s Depth API to get the distance to objects
• RoboflowCaller - Entire Application logic
• BuildingBlocks - Meta XR blocks for easy integration & project setup.
• HandActivator - Turn on Hand GUI if watching on inner hand

• Internet Connection - Ensure your app has a secure and stable internet connection.
• IP Address - Verify that the IP address in your “RoboflowCaller” object is correct.
• Local Inference - Make sure your Docker and Roboflow Local Inference are running. The first time you start, the model may take a few minutes to download.
• Poker Odds Server - Confirm that the Node.js Poker Odds Server is active and running.
• Permissions - On Meta Quest, double-check that all necessary permissions are enabled in the settings.

• SensAI Kits GitHub - Main hub for all XR AI kits
• SensAI Hackademy - Early access program for courses and toolkits
• SensAI Hack - Upcoming hackathons where you can use the kits

PokerAssets: https://github.com/zardtomcat
Icons: Freepik - Flaticon

MIT – Free to use, modify and learn from.