ClassificationService class

Service for handling mosquito species classification using PyTorch models.

This service provides functionality to load a pre-trained mosquito classification model and classify images to identify mosquito species. It handles model loading, image preprocessing, and prediction results using on-device machine learning.

Usage Example

final pytorchWrapper = PytorchWrapper();
final classificationService = ClassificationService(
  pytorchWrapper: pytorchWrapper,
);

// Load the model (required before classification)
await classificationService.loadModel();

// Classify an image
final imageFile = File('/path/to/mosquito_image.jpg');
final result = await classificationService.classifyImage(imageFile);
print('Species: ${result['scientificName']}');
print('Confidence: ${result['confidence']}');

Model Architecture

The service uses a PyTorch Lite model optimized for mobile deployment:

• Model: Pre-trained CNN for mosquito species classification
• Input: 224x224 RGB images with ImageNet normalization
• Output: Species probabilities across trained classes
• Performance: Optimized for real-time inference on mobile devices

Supported Platforms

• ✅ Android (API 21+)
• ✅ iOS (iOS 12.0+)
• ❌ Web (PyTorch Lite not supported)
• ❌ Desktop (PyTorch Lite not supported)

Error Handling

The service handles various error conditions:

• Platform not supported (throws PlatformException)
• Model loading failures (throws Exception)
• Invalid image formats (handled by PyTorch Lite)
• Memory constraints (handled by PyTorch Lite)

See also:

• ClassificationRepository for higher-level classification operations
• PytorchWrapper for the underlying PyTorch integration
• ClassificationModel for the model interface