Real-Time AI Video Analytics Dashboard

This project is a real-time video analytics dashboard using YOLOv10 and LLM (Large Language Model) integration, built with Flask and Ultralytics YOLO. It provides live object detection, analytics, and AI-powered insights from any video source (webcam, IP camera, or video file).

Images

Features

• Live Video Feed: View real-time video from webcam, IP stream, or video file.
• Object Detection: Uses YOLOv10 for detecting people, vehicles, weapons, and more.
• Analytics Panels: Live stats for queue, crowd, weapons, traffic, and parking.
• Detection History: Per-second detection graphs for all features.
• LLM Insights: AI-generated summaries and Q&A about the current video feed and analytics.
• Alerts: Real-time alerts for crowding, weapons, and traffic events.
• Modern Dashboard UI: Responsive, dark-mode enabled, with sidebar navigation.

Getting Started

Prerequisites

• Python 3.8+
• pip
• (Recommended) A CUDA-capable GPU for best YOLO performance

Installation

1. Clone the repository

   git clone <your-repo-url>
   cd InovHack

2. Install dependencies

   pip install -r requirements.txt

   (If requirements.txt is missing, install manually:)

   pip install flask opencv-python ultralytics torch requests

3. Download YOLOv10 model weights

   • Download yolov10n.pt from the Ultralytics YOLOv10 release page and place it in the project root.

4. Set your OpenRouter API key

   • Get a free API key from OpenRouter.
   • Set it as an environment variable:

     set OPENROUTER_API_KEY=sk-or-...

Running the App

python app.py

• The dashboard will be available at http://127.0.0.1:5000

Usage

• Switch Video Source: Use the dropdown above the video feed to select webcam (Live Camera) or enter a video/IP stream URL.
• View Analytics: Use the sidebar to switch between Queue, Crowd, Weapons, Traffic, and Parking panels.
• Ask the AI: Use the LLM Insights panel to ask questions about the current analytics.

File Structure

• app.py - Main Flask backend, YOLO/LLM logic, API endpoints
• templates/dashboard.html - Dashboard UI (Bootstrap 5, Chart.js)
• yolov8n.pt or yolov10n.pt - YOLO model weights
• requirements.txt - Python dependencies

Notes

• For webcam, ensure your device has a camera and OpenCV can access it (index 0).
• For video files or IP streams, enter the full path or URL in the dashboard.
• LLM features require a valid OpenRouter API key and internet access.

License

MIT License

---

Made for InovHack 2.0 by DevSidd2006