Quest 3 AR Vision - AI Object Detection

An interactive WebXR (AR) application designed for Meta Quest 3 and compatible AR devices. This project combines real-time AI object detection with augmented reality to bridge the gap between the physical and digital worlds.

🚀 Features

• AI Object Detection: Uses TensorFlow.js and the COCO-SSD model to identify real-world objects (people, chairs, laptops, phones, etc.) directly in your AR view.
• Surface Hit-Test: Detects surfaces and allows you to place virtual 3D objects (Torus Knots) in your physical space.
• Plane Detection: Automatically identifies floors, walls, and tables, highlighting them with semi-transparent overlays.
• Modern UI: A premium user interface with glassmorphism effects, a welcome screen, and real-time status updates.
• WebXR Integration: Built on top of the WebXR Device API for high-performance AR experiences in the browser.

🛠️ Tech Stack

• Three.js: For 3D rendering and XR session management.
• TensorFlow.js: For running machine learning models in the browser.
• COCO-SSD Model: Pre-trained object detection model.
• WebXR API: For accessing AR features on compatible headsets and mobile devices.

🏁 Getting Started

Prerequisites

• A WebXR-compatible browser (e.g., Meta Quest Browser, Google Chrome on Android).
• A device with AR capabilities (Meta Quest 3, Meta Quest Pro, or ARCore-compatible Android devices).
• HTTPS connection (required for WebXR and camera access).

Running Locally

1. Clone the repository:

   git clone https://github.com/bciric1/ARVR.git

2. Serve the directory using a local HTTPS server. You can use serve or any similar tool:

   npx serve

   Note: For Meta Quest, you may need to use a tool like ngrok to provide a secure HTTPS tunnel if you're not hosting on a remote server.

🎮 How to Use

1. Open the application in your VR/AR browser.
2. Click "Započni iskustvo" (Start Experience) on the welcome screen.
3. Wait for the AI model to load (status will update to "AI model ucitan!").
4. Click "Enter AR".
5. AI Detection: Look around; the AI will periodically scan and label objects it recognizes.
6. Placing Objects: Aim the green ring cursor at a surface and pull the trigger (or click) to place a 3D object.

📝 License

This project is open-source and available under the MIT License.