Camera Analyst - AI-Powered F&B Management System

Overview

Camera Analyst is an advanced Artificial Intelligence system designed for the Food & Beverage (F&B) industry. It transforms traditional surveillance cameras into active business intelligence tools. By analyzing video feeds in real-time, the system provides actionable insights regarding staff performance, food quality control, and customer behavior, helping restaurant owners optimize operations and reduce costs.

Key Features

The system consists of several core AI modules:

1. AI Biometric Attendance: Automated staff check-in/check-out using facial recognition, capable of adapting to appearance changes over time.
2. Human Action Recognition (HAR): Analyzes staff movements to distinguish between active working time and idle time, providing a true measure of labor efficiency.
3. Food Quality Matching: "Digital Chef" that compares outgoing dishes against master images to ensure consistency in presentation and portioning before they reach the customer.
4. Customer Flow Analytics: Counts customers, tracks dwell time, and analyzes service speed to improve customer experience.
5. Automated Reporting: Consolidates data into actionable financial and operational reports.

Web Interface Modules

The application includes the following dashboards and interfaces (located in app_s/):

• Authentication: login.html, admin_login
• Executive Dashboard: dashboard.html, executive_dashboard_desktop
• Live Monitoring: live-view.html, live_camera_streaming_grid_view
• Staff Management:
  • Attendance: staff-attendance.html, staff_attendance_report_desktop
  • Performance: staff-performance.html, staff_performance_desktop
  • Receptionist: receptionist-dashboard.html, receptionist_performance_dashboard
• Operational Analytics:
  • Kitchen Traffic: kitchen-analytics.html, kitchen_traffic_analytics
  • Food QC: food-qc.html, food_qc_monitor_desktop
  • Customer Flow: customer-analytics.html, customer_flow_analytics_desktop
• System Configuration:
  • Camera/RTSP Setup: rtsp-setup.html, camera_configuration_rtsp_setup
  • AI Configuration: camera_ai_configuration

Getting Started

Prerequisites

• Python 3.10 higher

DeepStream Project Structure

This directory contains the template for a DeepStream AI application.

Directory Structure

• configs/: Configuration files for DeepStream elements (GIE, Tracker, Sources).
  • deepstream_config.txt: Main application configuration file.
• src/: Source code for the application.
  • main.py: Main entry point for the Python application.
  • api/: API endpoints for controlling the application (FastAPI/Flask).
  • core/: Core logic for pipeline management and inference handling.
• scripts/: Helper scripts for deployment and management.

Prerequisites

• NVIDIA Jetson or dGPU system
• DeepStream SDK installed (version 6.x or later recommended)
• Python 3.x
• GStreamer plugins

Usage

1. Navigate to src/.
2. Run the main application:

   python3 main.py

Customization

• Edit configs/deepstream_config.txt to add/remove sources or change model paths.
• Modify src/main.py to implement custom logic for metadata processing or integration with external systems.

Installation & Running

1. Clone the repository or download the project files.

2. Navigate to the project root directory.

3. Run the included server script:

   python3 run_server.py

4. Open your web browser and navigate to:

   • Login: http://localhost:8000/login.html
   • Dashboard: http://localhost:8000/dashboard.html

Project Structure

AI_camera/
├── app/                  # Web application source files (HTML, assets)
│   ├── admin_login/
│   ├── camera_ai_configuration/
│   ├── customer_flow_analytics_desktop/
│   ├── ... (other modules)
│   ├── *.html              # Main interface pages
├── run_server.py           # Python script to host the web app locally
├── camera_analyst.md       # Detailed project documentation and specs
├── skills.md               # Technical skills and architecture notes
└── README.md               # This file

Security & Privacy

• Edge Processing: Video data is processed locally at the restaurant (Edge Server) to ensure privacy.
• Data Minimization: Only metadata and statistical reports are sent to the cloud.
• Face Vectorization: Facial data is stored as mathematical vectors, not images.

Contact

DSC-Labs - [Contact Information]