PRODUCTION RELEASE: v1.0.0Active Live Deployment

Real-Time Edge Computer Vision &
Spatial Foot-Traffic Analytics

VisionEye delivers sub-12ms multi-person detection, ByteTrack spatial tracking, interactive 4-point homography 2D minimaps, and automated IN/OUT counting on iOS, Android, macOS, Windows, and Linux.

Launch Live Web Application Instant Access • No Installation Required
Universal Web & Mobile Support 100% Free & Open Source Works on iPhone, Android & Desktop
VisionEye Architecture & System Design

All Supported Platforms & Devices

Launch instantly in any browser or download native standalone packages for macOS, Windows, or Linux.

📱

iPhone, iPad & Android

Smartphones, Tablets & Mobile Browsers

  • ✓ Zero install needed — runs instantly in Safari / Chrome
  • ✓ Live camera streaming with front & rear lens switching
  • ✓ Touch-optimized real-time spatial telemetry HUD
⚡ Launch Mobile Web App
🍏

macOS (Apple Silicon)

M1 / M2 / M3 / M4 • macOS 12 Monterey or newer

  • ✓ Hardware CoreML Neural Acceleration
  • ✓ Ultra-low 9ms inference latency
  • ✓ Standalone desktop package
Download for Mac (Apple Silicon)
🍏

macOS (Intel x64)

Intel Core i5/i7/i9 • macOS 11 Big Sur or newer

  • ✓ Multi-threaded CPU ONNX Runtime
  • ✓ Integrated FaceTime HD camera support
  • ✓ Standalone desktop package
Download for Mac (Intel)
🪟

Windows (64-bit)

Windows 10 / 11 (64-bit)

  • ✓ DirectShow & USB Camera drivers
  • ✓ One-click setup wizard
  • ✓ High-DPI display optimized HUD
Download for Windows (.exe)
🐧

Linux (x86_64)

Ubuntu, Debian, Fedora, Arch, etc.

  • ✓ Portable standalone AppImage
  • ✓ Standard glibc 2.31+ compatibility
  • ✓ Edge AI ready
Download for Linux (AppImage)

End-to-End Modular Architecture

How VisionEye transforms high-definition video into real-time spatial intelligence across 5 interconnected architectural tiers.

VisionEye System Architecture Flowchart
T1

1. Ingestion & Input Streams

Ingests 30–60 FPS video streams from phone cameras, USB/FaceTime HD webcams, RTSP surveillance cameras, and MP4/MOV video files with seamless looping.

T2

2. Edge CV & Neural Pipeline

Applies 640×640 letterbox scaling, ultra-fast YOLO26n ONNX neural inference (9–12ms), ByteTrack Kalman state tracking, virtual counting line debouncing, and 4-point homography 2D ground projection.

T3

3. High-Speed Telemetry Server

Asynchronous streaming backend broadcasting sub-millisecond telemetry updates via WebSockets, serving high-speed MJPEG video feeds, and coordinating analytics state.

T4

4. Universal Client Layer

Responsive React web application engineered for flawless operation on iPhone, Android smartphones, iPads, tablets, laptops, and desktop computers.

T5

5. Cloud & Production Infrastructure

Containerized production deployment on Microsoft Azure with Let's Encrypt TLS/SSL encryption, high-concurrency Nginx reverse proxy, and 24/7 resilience.

Engineered for Real-Time Edge Intelligence

Zero external cloud dependencies. Everything runs locally on your workstation or directly in your browser.

YOLO26n ONNX Neural Engine

Ultra-lightweight person detection model delivering 9ms inference latency with hardware neural acceleration.

ByteTrack Multi-Person Tracking

Industrial Kalman filtering with Hungarian association maintains persistent track IDs across dense pedestrian crowds and brief occlusions.

4-Point Homography Top-View Minimap

Transform angled surveillance camera footage into a 2D bird's-eye orthographic ground plane with 1-click auto-calibration and real-time perspective mapping.

Virtual Counting Line & Analytics

Sub-pixel line-crossing detection with directional enter/exit debouncing, real-time venue occupancy tracking, and timeline flow history.

Privacy Protection Masking

On-the-fly anonymization applying dynamic Gaussian blur or pixelation filters over detected human bounding boxes prior to stream encoding.

Universal Video Input Feeds

Supports smartphone cameras (front/rear), webcams, MP4/MOV video files (with continuous looping), and synthetic high-density streams.

Simple Setup Guide

Launch instantly in your browser or set up native application packages on your computer.

  • 1

    Instant Access on Any Device

    No app store downloads or installations required! VisionEye runs natively inside Safari, Google Chrome, Firefox, Edge, and Samsung Internet on iPhone, iPad, Android smartphones, and tablets.

  • 2

    Launch the Live Application

    Tap the button below to open the real-time AI foot-traffic analytics dashboard directly in your browser:

  • 3

    (Optional) Add to Your Mobile Home Screen

    You can turn VisionEye into a full-screen standalone app icon on your phone:

    🍏 On iPhone/iPad (Safari): Tap the Share button (⎋ with arrow) at bottom ➔ Tap "Add to Home Screen".
    🤖 On Android (Chrome): Tap the 3 dots menu (⋮) at top right ➔ Tap "Add to Home screen" or "Install app".
  • 1

    Download the Installer for Mac

    Choose the installer package tailored for Apple Silicon Macs (M1, M2, M3, M4) or Intel-based Mac computers.

  • 2

    Open Installer & Drag to Applications

    Open the downloaded package, drag the VisionEye icon and drop it into your Applications folder.

  • 3

    Launch and Start Tracking

    Open VisionEye from your Applications folder or Spotlight Search. Grant camera permissions or upload any video clip to begin real-time foot-traffic analytics.

    💡 First Launch on macOS: If prompted on first launch, right-click on VisionEye in Applications and select Open to confirm.
  • 1

    Download the Windows Setup Package

    Download the Windows installer package from the downloads section above.

  • 2

    Run the Setup Wizard

    Double-click the installer to start the step-by-step setup wizard.

    💡 Windows SmartScreen Tip: If prompted with "Windows protected your PC", click "More info" and select "Run anyway".
  • 3

    Launch & Analyze

    Launch VisionEye from your Desktop or Start Menu. Connect your webcam or upload video clips to view real-time occupancy and IN/OUT counts.

  • 1

    Download the Linux Package

    Download the standalone VisionEye Linux package for your distribution.

  • 2

    Enable Execution Permission

    Right-click the downloaded file, select Properties ➔ Permissions, and check "Allow executing file as program".

  • 3

    Launch Application

    Double-click the package to launch VisionEye instantly on Ubuntu, Debian, Fedora, Arch, and compatible Linux distributions.

System Requirements

Hardware and operating system prerequisites for running VisionEye at 30 FPS.

Component Minimum Requirements Recommended (Optimal 30+ FPS)
Operating System iOS 15+, Android 10+, macOS 12+, Windows 10, Ubuntu 20.04+ iOS 17+, Android 14+, macOS 14 Sonoma, Windows 11
Processor Standard Mobile SoC / Quad-core Intel / AMD / Apple M1 Apple Silicon M2/M3/M4 or Multi-Core Intel Core i7
Memory (RAM) 3 GB RAM 8 GB RAM or higher
Disk Space Zero for Web App (500 MB for Desktop) 1 GB SSD storage for local video recordings
Video Capture Smartphone camera or standard 720p USB webcam 1080p HD camera / RTSP IP stream

Production Release Updates

VisionEye v1.0.0 — Production Cloud Release

Live 24/7 Deployment
  • Mobile Camera Integration: Real-time smartphone front & rear camera streaming with live AI bounding boxes.
  • 1-Click Auto-Calibrate: Instant ground floor perspective fitting with camera angle presets.
  • 🔄 Continuous Video Looping: Seamless EOF auto-reopen playback without stream freezing.
  • 📊 Top-View 2D Radar: Orthographic bird's-eye spatial positioning and trajectory breadcrumbs.
  • 🛡️ Privacy Protection: Real-time Gaussian blur and pixelation filters.
  • 🌐 24/7 Cloud Architecture: Automated HTTPS SSL reverse proxy hosted on pvsairamsaketh.in.