ML Vision - Object Detection App

A Flutter application that uses Google ML Kit to detect and classify objects in photos captured with the device camera. Built for iOS 15.5+ and Android, this app demonstrates powerful on-device machine learning capabilities.

Features

• Camera Integration: Tap the camera button to capture photos using the device camera
• Real-time Object Detection: Leverages Google ML Kit for accurate object detection and classification
• Visual Bounding Boxes: Displays color-coded boxes around detected objects in the image
• Interactive Selection: Tap any detected object card to highlight its corresponding bounding box
• Detailed Results: Displays detected objects with:
  • Primary object label with highest confidence
  • Multiple classification possibilities per object
  • Confidence scores (percentage)
  • Visual confidence indicators (HIGH/MEDIUM/LOW)
  • Color-coded badges
  • Bounding box coordinates
• Adjustable Settings: Configure detection parameters:
  • Confidence threshold slider (adjust sensitivity)
  • Image quality control
• Split-Screen Layout: Fixed image display at top with scrollable object list below
• Clean UI: Modern Material Design 3 interface
• On-device Processing: All detection happens locally using Google's ML Kit
• Multi-label Classification: Shows alternative classifications for each detected object

Requirements

• iOS: 15.5 or later (compatible with iOS 26)
• Android: API Level 21 or later
• Xcode: 13.0 or later (for iOS development)
• Flutter: 3.9.2 or later
• CocoaPods: Latest version (for iOS)

Technologies Used

• Flutter: Cross-platform framework
• Google ML Kit: On-device machine learning for object detection and image labeling
• Packages:
  • google_mlkit_object_detection ^0.15.0 - Google ML Kit object detection
  • google_mlkit_image_labeling ^0.14.1 - Google ML Kit image labeling
  • image_picker ^1.1.2 - Camera and photo library access

Getting Started

1. Clone and Setup

cd ml_vision
flutter pub get

2. Install iOS Dependencies

cd ios
pod install
cd ..

3. Run the App

Connect your iOS device or start the iOS simulator:

flutter run

Or open in Xcode:

open ios/Runner.xcworkspace

iOS Configuration

The project includes all necessary iOS configurations:

Deployment Target

• Set to iOS 15.5 in both Podfile and Xcode project settings

Permissions

The following permissions are configured in Info.plist:

• NSCameraUsageDescription: Camera access for taking photos
• NSPhotoLibraryUsageDescription: Photo library access for image selection

Bundle Identifier

• com.example.mlVision

How to Use

1. Launch the app on your iOS device
2. Tap the camera button (floating action button) to open the camera
3. Take a picture of objects you want to detect
4. View results showing:
   • Captured image
   • List of detected objects
   • Confidence scores for each detection

Project Structure

lib/
└── main.dart              # Main app with object detection logic

ios/
├── Podfile                # CocoaPods dependencies (iOS 15.5)
├── Runner.xcodeproj/      # Xcode project
└── Runner/
    └── Info.plist         # iOS permissions and configuration

Troubleshooting

ML Kit dependencies not found

Run:

flutter clean
flutter pub get
cd ios
pod install
cd ..

Building for iOS-13.0 error

Ensure the deployment target is set to iOS 15.5:

• Check ios/Podfile: platform :ios, '15.5'
• Check Xcode project settings: Target → Deployment Info → iOS 15.5

No objects detected

ML Kit works best with:

• Clear, well-lit images
• Distinct objects with defined shapes
• Common everyday objects (furniture, food, animals, vehicles, etc.)
• Objects that aren't too small in the frame

Camera not working

Verify camera permissions are granted in device Settings → Your App → Camera

Known Limitations

• Requires physical device with camera for best results
• Object detection accuracy depends on Google ML Kit models
• Detection works best with common objects in well-lit conditions
• May not detect very small objects or unusual items

License

This project is a demonstration application for educational purposes.

Acknowledgments

• Built with Flutter
• Uses Google ML Kit
• Object detection via google_mlkit_object_detection package