Pool Area Occupancy AI

Web-based AI system analyzing resorts' pool areas. It detects people and chairs, and calculates occupancy. It also provides insights (e.g., demographics) with a multi-modal LLM.

The project consists of:

• Frontend: A barebone interface supporting file uploads and gallery image selection.
• Backend: Provides statistics and insights using YOLO, Gemini, and some geometric processing.

Prerequisites

1. Docker: Install Docker and Docker Compose.
2. YOLO: The code downloads YOLO11x (i.e., the latest at the time of writing).
3. Gemini API Key: Add your API key to your environment.

How to use

1. Clone the repository:

   git clone https://github.com/your-repo/chair-occupancy-ai.git
   cd chair-occupancy-ai

2. Add your Gemini API key to the .env file:

   echo "GEMINI_API_KEY=your_api_key_here" > .env

3. Start the application:

   ./run.sh

4. Access the frontend at localhost.

5. Upload an image or select one from the gallery.

6. Alternatively, you can access the backend API directly at /docs for testing.

API Endpoints

• Analyze Image: POST /analyze-image

  • Input: Image file.
  • Output: Detection results and labeled image.

• AI Analysis: POST /llm-analyze

  • Input: Image file.
  • Output: GenAI insights.

Tech Stack

• AI Models: YOLO, Google Gemini
• Backend: FastAPI, Python
• Frontend: HTML, CSS, JavaScript
• Docker: For containerization