Eyes

Intelligent Video Surveillance for macOS and iOS

Monitor your IP cameras with advanced AI-powered detection. Person, vehicle, and animal detection, facial recognition, and anomaly detection. Accessible from your Mac, iPhone, and iPad.

macOS iOS / iPadOS SwiftUI Core ML Metal

Features

📹

Multi-Camera

Manage multiple IP cameras simultaneously with RTSP and ONVIF support. Grid display with real-time video streaming via high-performance Metal rendering.

🧠

AI Detection

Intelligent detection of people, vehicles, and animals using Apple Vision framework, YOLO, and MobileNetV2-SSDLite models. Configurable sensitivity and minimum size per object type.

👤

Facial Recognition

Face detection and recognition with custom CreateML model. Personalized training to identify known individuals.

⚠️

Anomaly Detection

Anomaly detection based on DINOv3 with automatic baseline learning. Identifies unusual events without prior configuration.

🎯

Object Tracking

Track objects between frames with ByteTrack or Re-ID (OSNet). Follow movements of people, vehicles, and animals across frames.

📼

Smart Recording

Recording triggered by configurable rules (AND/OR). Pre-buffer and post-buffer to capture context. Fragmented or standard MP4 format.

🔍

Classification Zones

Define zones on the image to trigger specific actions when an object enters. Custom classification per zone.

📊

Interactive Timeline

View recordings on a timeline with smart thumbnails. Keyboard navigation, pinch zoom, automatic grouping.

🌐

Secure Web Server

Access your cameras and recordings via a secure web interface with authentication. REST API for integration with other systems and the mobile app.

📱

Mobile App

View your cameras and recordings from your iPhone or iPad. Native iOS interface optimized for all screen sizes.

🎵

Audio

RTSP audio stream support with volume control. Live listening from the main interface.

💾

Automatic Retention

Automatic disk space management with deletion of old recordings based on configurable retention period per camera.

Screenshots - macOS

Eyes main interface - Camera grid
Main Interface

Multi-camera grid view with real-time detection. Recent events displayed at the bottom with clickable thumbnails.

Eyes recordings window
Recordings Window

Multi-camera timeline with temporal navigation. Filter by detection type, interactive zoom, and integrated video playback.

Mobile App - iOS / iPadOS

EyesMobile lets you view your cameras and recordings from anywhere. The app connects to the Eyes server on your Mac via the built-in REST API.

EyesMobile app on iPhone
EyesMobile

Live stream viewing and recording navigation. Native iOS interface with quick camera and workspace selection.

Technology Stack

Eyes is a native macOS application built with Swift and SwiftUI. It leverages cutting-edge technologies for optimal performance and intelligent detection.

Apple Frameworks

  • SwiftUI - Modern and reactive user interface
  • Core ML - Machine learning model inference
  • Vision - Object and face detection
  • Metal - High-performance video rendering
  • AVFoundation - Video playback and recording
  • VideoToolbox - Hardware H.264/H.265 decoding
  • CreateML - Custom model training

External Dependencies

FFmpegKit

FFmpeg libraries compiled for macOS with VideoToolbox hardware acceleration, enabling RTSP stream decoding and advanced video manipulation.

  • Repository: github.com/arthenica/ffmpeg-kit
  • License: LGPL v3.0
  • Components used: libavcodec, libavformat, libavutil, libswscale, libswresample, libavfilter, libavdevice

Swift NIO

Asynchronous non-blocking network framework for the built-in web server and HTTP communications.

AI Models

  • Apple Vision - Built-in person, face, and animal detection
  • YOLO - Real-time object detection (people, vehicles, animals)
  • MobileNetV2-SSDLite - Lightweight and fast object detection
  • DINOv3 ViT-S/16 - Feature extraction for anomaly detection
  • OSNet - Object re-identification for cross-frame tracking
  • CreateML Classifier - Custom facial classification

Supported Protocols

  • RTSP - Real Time Streaming Protocol (TCP/UDP)
  • ONVIF - Open Network Video Interface Forum
  • H.264/AVC - Standard video codec
  • H.265/HEVC - High efficiency video codec

Architecture

  • POSIX Structure - Unix-compatible recording storage
  • REST API - Server-client communication for mobile app
  • Multi-threading - Optimized parallel processing per camera
  • Smart Cache - Optimized memory management for thumbnails and models

Licenses

FFmpeg / FFmpegKit

FFmpegKit (by arthenica) provides FFmpeg libraries compiled with LGPL v3.0 license and VideoToolbox hardware acceleration.

By using Eyes, you accept the terms of the LGPL v3.0 license for FFmpeg components.

Read LGPL v3.0 License

Swift NIO

Swift NIO is developed by Apple and distributed under the Apache 2.0 license.

Read Apache 2.0 License

YOLO (Ultralytics)

YOLO object detection models are developed by Ultralytics and distributed under the AGPL-3.0 license.

Read AGPL-3.0 License

DINOv3 (Meta AI)

DINOv3 vision transformer models are developed by Meta AI Research and distributed under the Apache 2.0 license.

Read Apache 2.0 License

MobileNetV2-SSDLite

MobileNetV2-SSDLite is a lightweight object detection model originally developed by Google and distributed under the Apache 2.0 license.

Read Apache 2.0 License

OSNet

OSNet (Omni-Scale Network) for person re-identification is distributed under the MIT license.

Read MIT License

System Requirements

Eyes (macOS)

  • System: macOS 14.0 (Sonoma) or later
  • Processor: Apple Silicon (M1/M2/M3/M4) or Intel with Neural Engine
  • RAM: 8 GB minimum, 16 GB recommended for multiple cameras
  • Storage: Variable depending on recording retention
  • Network: Stable connection to IP cameras

EyesMobile (iOS/iPadOS)

  • System: iOS 17.0 / iPadOS 17.0 or later
  • Devices: iPhone, iPad
  • Network: Connection to Eyes server (local or remote)

Contact

For any questions, suggestions, or bug reports, contact us:

info@pghsoftware.be

© 2025 PGH Software. All rights reserved.