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 instantly in any browser or download native standalone packages for macOS, Windows, or Linux.
How VisionEye transforms high-definition video into real-time spatial intelligence across 5 interconnected architectural tiers.
Ingests 30–60 FPS video streams from phone cameras, USB/FaceTime HD webcams, RTSP surveillance cameras, and MP4/MOV video files with seamless looping.
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.
Asynchronous streaming backend broadcasting sub-millisecond telemetry updates via WebSockets, serving high-speed MJPEG video feeds, and coordinating analytics state.
Responsive React web application engineered for flawless operation on iPhone, Android smartphones, iPads, tablets, laptops, and desktop computers.
Containerized production deployment on Microsoft Azure with Let's Encrypt TLS/SSL encryption, high-concurrency Nginx reverse proxy, and 24/7 resilience.
Zero external cloud dependencies. Everything runs locally on your workstation or directly in your browser.
Ultra-lightweight person detection model delivering 9ms inference latency with hardware neural acceleration.
Industrial Kalman filtering with Hungarian association maintains persistent track IDs across dense pedestrian crowds and brief occlusions.
Transform angled surveillance camera footage into a 2D bird's-eye orthographic ground plane with 1-click auto-calibration and real-time perspective mapping.
Sub-pixel line-crossing detection with directional enter/exit debouncing, real-time venue occupancy tracking, and timeline flow history.
On-the-fly anonymization applying dynamic Gaussian blur or pixelation filters over detected human bounding boxes prior to stream encoding.
Supports smartphone cameras (front/rear), webcams, MP4/MOV video files (with continuous looping), and synthetic high-density streams.
Launch instantly in your browser or set up native application packages on your computer.
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.
Tap the button below to open the real-time AI foot-traffic analytics dashboard directly in your browser:
You can turn VisionEye into a full-screen standalone app icon on your phone:
Choose the installer package tailored for Apple Silicon Macs (M1, M2, M3, M4) or Intel-based Mac computers.
Open the downloaded package, drag the VisionEye icon and drop it into your Applications folder.
Open VisionEye from your Applications folder or Spotlight Search. Grant camera permissions or upload any video clip to begin real-time foot-traffic analytics.
Download the Windows installer package from the downloads section above.
Double-click the installer to start the step-by-step setup wizard.
Launch VisionEye from your Desktop or Start Menu. Connect your webcam or upload video clips to view real-time occupancy and IN/OUT counts.
Download the standalone VisionEye Linux package for your distribution.
Right-click the downloaded file, select Properties ➔ Permissions, and check "Allow executing file as program".
Double-click the package to launch VisionEye instantly on Ubuntu, Debian, Fedora, Arch, and compatible Linux distributions.
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 |