OpenCount

AI-powered object counting for every platform. Native. Private. Open source.

OpenCount is a free, open-source, cross-platform AI-powered object counter that surpasses commercial alternatives (CountThings, Zap Count) with on-device AI detection, voice counting, AR counting, collaboration, and extensive export capabilities.

Architecture

OpenCount/
├── apps/
│   ├── ios/                 # iOS/iPadOS native app (SwiftUI)
│   │   ├── app/             # Main target
│   │   ├── watch/           # watchOS companion
│   │   ├── widget/          # WidgetKit extension
│   │   ├── clip/            # App Clip
│   │   ├── tests/           # Property-based tests (SwiftCheck)
│   │   └── uitests/         # XCUITest smoke tests
│   ├── android/             # Android native app (Jetpack Compose)
│   ├── desktop/             # Desktop app (Compose for Desktop - Win/Mac/Linux)
│   └── web/                 # Web app (Kotlin/JS, scaffold)
├── packages/
│   └── shared/              # KMP shared module (cross-platform core)
│       └── src/
│           ├── commonMain/  # Shared business logic
│           ├── iosMain/     # iOS-specific implementations
│           ├── androidMain/ # Android-specific implementations
│           ├── desktopMain/ # Desktop-specific implementations
│           └── jsMain/      # Web-specific implementations
├── docs/                    # Documentation
├── project.yml              # XcodeGen project spec (iOS)
├── build.gradle.kts         # Gradle root (Android/Desktop/Web)
└── settings.gradle.kts      # Gradle settings

Platform Coverage

Platform	Tech Stack	Status
iOS 16+ / iPadOS	SwiftUI, Vision, CoreML, ARKit	✅ Stable
watchOS 9+	SwiftUI, WCSession	✅ Stable
macOS (Mac Catalyst)	SwiftUI (shared iOS code)	🔄 Planned
Android	Kotlin, Jetpack Compose	✅ Stable (APK ~10MB)
Desktop (Win/Mac/Linux)	Kotlin, Compose for Desktop	✅ Stable
Web	Kotlin/JS	🚧 Interactive prototype

Features

• AI-Powered Detection: On-device YOLOv8 via CoreML/Vision (iOS), ML Kit (Android)
• Manual Counting: Tap-to-count with multi-type categorization
• Voice Counting: Speech-to-text hands-free counting (multi-language)
• Live Camera Counting: Real-time object detection
• AR Counting: Augmented reality overlay counting (iOS ARKit)
• Video Counting: Frame-by-frame video analysis
• Multi-Image Sessions: Organize multiple images per session
• Regions: Define rectangular, elliptical, or polygonal count zones
• Formulas: Custom expressions for derived metrics
• Tags & Search: Color-coded tags with full-text search
• Collaboration: Real-time sync via CloudKit
• Apple Pencil / Stylus: Precision marking with hover ghost
• Tally Mode: Distraction-free counting interface
• Heatmap: Density visualization of markers
• Export: CSV, XLSX, JSON, COCO JSON, PDF, Annotated Image
• Bulk Export: Multi-session ZIP export
• i18n: 8 languages (English, Vietnamese, Japanese, Korean, Chinese, French, German, Spanish)
• Templates: Private template library + public marketplace

AI Engine

• On-device: All AI runs locally — zero data leaves the device
• Default Model: YOLOv8-nano via CoreML (iOS) / TensorFlow Lite (Android)
• Custom Models: Import CoreML (.mlpackage) / TFLite models
• Smart Counting: Duplicate detection, clustering, fatigue warnings
• Panorama Support: Auto-tiling for large images with NMS dedup

Quick Start

Prerequisites

• iOS: Xcode 16+, CocoaPods or SPM (automatic)
• Android/Desktop: JDK 17+, Gradle 8.9+
• All Platforms: Git

iOS

make ios-generate     # Generate Xcode project
make ios-build        # Build for simulator
make ios-test         # Run iOS tests
make ios-ipa          # Build unsigned IPA

Android, Desktop & Shared (KMP)

./gradlew :packages:shared:check   # Build + test shared module
./gradlew :apps:android:assembleDebug  # Build Android APK
./gradlew :apps:desktop:run            # Run Desktop app

All Tests

# KMP (cross-platform) — 200+ tests
./gradlew :packages:shared:desktopTest

# Android build
./gradlew :apps:android:assembleDebug

# Desktop build
./gradlew :apps:desktop:compileKotlinDesktop

# Web build
./gradlew :apps:web:compileKotlinJs

# iOS (requires macOS)
make ios-test

Testing Strategy

Layer	Type	Framework	Location
Unit	Model & utility tests	kotlin.test	packages/shared/src/commonTest/
Integration	Cross-service workflows	kotlin.test	packages/shared/src/commonTest/
E2E	Full user workflow simulation	kotlin.test	packages/shared/src/commonTest/e2e/
Thread Safety	Concurrent access & race conditions	kotlin.test	packages/shared/src/commonTest/
Stress	Large datasets (10K+ markers)	kotlin.test	packages/shared/src/commonTest/
Property	Invariant verification	kotlin.test	packages/shared/src/commonTest/
i18n	Multi-language string validation (8 langs)	kotlin.test	packages/shared/src/commonTest/i18n/
SwiftCheck	iOS property invariants	SwiftCheck	apps/ios/tests/
UI	XCUITest smoke & accessibility	XCTest	apps/ios/uitests/

Internationalization

8 languages built into the shared KMP module:

• English (default), Vietnamese, Japanese, Korean, Chinese
• French, German, Spanish

Switch at runtime via Strings.language = "vi".

License

Open source. Free forever. No telemetry, no accounts required.

Comparison

Feature	OpenCount	CountThings	Zap Count
Native iOS	✅ SwiftUI	✅	✅
Native Android	🚧 Jetpack Compose	❌	❌
Desktop App	🚧 Compose Desktop	❌	❌
Web App	🗺️ Planned	❌	❌
On-device AI	✅ YOLOv8	❌ (cloud)	❌ (cloud)
Voice Counting	✅	❌	❌
AR Mode	✅ (ARKit)	❌	❌
Video Analysis	✅	❌	❌
Bulk Export	✅	❌	❌
Collaboration	✅ (CloudKit)	❌	❌
Custom AI Models	✅	❌	❌
Open Source	✅	❌	❌
Privacy (no cloud)	✅	❌	❌
Price	Free	$$$	$$$