Object Counter - Real-Time Object Detection for Arm

A native Android application that performs real-time object counting using computer vision, optimized for Arm-based mobile devices.

🎯 Project Overview

Object Counter is an on-device AI application that counts objects in real-time using your phone's camera or from photos in your gallery. Unlike cloud-based solutions, all processing happens locally on your Arm device, ensuring privacy, speed, and offline capability.

Why This Project Stands Out

• 100% On-Device Processing: No cloud dependencies, no data leaves your phone
• Adaptive Background Detection: Automatically identifies background color and detects objects that differ from it
• Watershed Algorithm: Accurately separates touching/overlapping objects
• Real-Time Performance: Optimized for Arm architecture with efficient OpenCV operations
• Dual Input Modes: Live camera detection and photo gallery analysis

✨ Features

Feature	Description
Live Detection	Real-time object counting through camera feed
Photo Analysis	Count objects from gallery images
Smart Separation	Watershed algorithm separates touching objects
Adaptive Detection	Auto-detects background color for accurate counting
Visual Feedback	Numbered bounding boxes overlay on detected objects
Count Stabilization	Smoothed counting reduces flickering in live mode

🛠️ Technical Implementation

Computer Vision Pipeline

1. Color Space Conversion: RGB → HSV for robust color analysis
2. Background Detection: Histogram-based dominant color identification
3. Binary Masking: Isolate objects from background
4. Morphological Operations: Noise reduction and gap filling
5. Distance Transform: Prepare for watershed segmentation
6. Watershed Algorithm: Separate touching objects
7. Contour Analysis: Extract bounding boxes and filter by size

Arm Optimization

• Uses OpenCV's optimized native libraries for Arm
• Efficient memory management with proper Mat cleanup
• Frame skipping strategy for smooth real-time performance
• YUV to RGB conversion optimized for camera pipeline

📱 Requirements

• Android 9.0 (API 28) or higher
• Arm-based Android device (arm64-v8a, armeabi-v7a)
• Camera permission

🚀 Build Instructions

Prerequisites

• Android Studio Hedgehog (2023.1.1) or newer
• JDK 17
• Android SDK 34

Step-by-Step Build

1. Clone the repository

   git clone https://github.com/Crawd-Counter/Crowd-Counter.git
   cd crowd-counter

2. Open in Android Studio

   • Launch Android Studio Hedgehog (2023.1.1) or newer
   • Select "Open an existing project"
   • Navigate to the cloned directory

3. Sync Gradle

   • Android Studio will automatically prompt to sync
   • Click "Sync Now" when prompted
   • Wait for dependencies to download (includes OpenCV)

4. Build the APK

   • Use Android Studio: Build → Build Bundle(s) / APK(s) → Build APK(s)
   • Or via terminal (after Android Studio generates wrapper):

     ./gradlew assembleDebug

5. Install on Device

   • Connect your Arm-based Android device via USB
   • Enable USB debugging in Developer Options
   • Click Run (▶️) in Android Studio

   Or via command line:

   adb install app/build/outputs/apk/debug/app-debug.apk

Running the App

1. Launch "Object Counter" from your app drawer
2. Grant camera permission when prompted
3. Live Mode: Tap "Detect" to start counting objects through camera
4. Photo Mode: Tap "Photo" to select an image from gallery
5. Tap "Reset" to return to camera view

📂 Project Structure

object-counter/
├── app/
│   ├── src/main/
│   │   ├── java/com/objectcounter/
│   │   │   ├── MainActivity.kt          # Main UI and camera handling
│   │   │   ├── ml/
│   │   │   │   └── ObjectDetector.kt    # Core detection algorithm
│   │   │   └── ui/
│   │   │       └── OverlayView.kt       # Detection visualization
│   │   ├── res/
│   │   │   ├── layout/                  # UI layouts
│   │   │   └── values/                  # Colors, themes
│   │   └── AndroidManifest.xml
│   └── build.gradle.kts
├── build.gradle.kts
├── settings.gradle.kts
├── LICENSE
└── README.md

🔧 Dependencies

• CameraX 1.3.1: Modern camera API for Android
• OpenCV 4.9.0: Computer vision library
• AndroidX: Core Android components
• Material Design: UI components

🎨 Use Cases

• Inventory Management: Count products, parts, or items
• Education: Count objects for learning activities
• Quality Control: Verify item counts in manufacturing
• Agriculture: Count seeds, fruits, or produce
• Research: Count specimens or samples

🏆 Hackathon Submission

This project was built for the Arm AI Developer Challenge 2025, demonstrating:

• ✅ AI/ML running locally on Arm architecture
• ✅ Innovative use of computer vision for practical problem-solving
• ✅ Production-ready code quality
• ✅ Optimized performance for mobile devices
• ✅ Open source with MIT license

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

---

Built with ❤️ for the Arm AI Developer Challenge 2025