Help & Documentation

Getting started is easy! Follow these steps:

1. Download and install the app from the App Store
2. Open the app and explore the free tier (3 datasets, 2 models, up to 50 images per inference session)
3. Import your first dataset from the Datasets tab (local folders, videos, Parquet, or DICOM files)
4. Import or select a model from the Models tab
5. Run inference on images from the Inference tab
6. Try live camera inference from the Camera tab
7. Upgrade to Premium for unlimited features and model training

Free Tier:

• 3 datasets maximum (lifetime total)
• 2 models maximum (lifetime total)
• Process up to 50 images per inference session
• Basic metrics and results
• All core features: classification, detection, segmentation, live camera

Premium:

• Unlimited datasets and models
• Model training (CreateML image classification)
• Export results (JSON, CSV, PDF)
• Advanced analytics and confusion matrices
• HuggingFace integration (download datasets and models)
• Unlimited training iterations
• Automation & trigger workflows powered by App Intents

You can import datasets from multiple sources in the Datasets tab:

• Local folders: Tap the \"+\" button and select a folder with images organized by class (e.g., \"cats/\", \"dogs/\"). The app automatically detects classes.
• Videos: Import video files and the app will extract frames automatically
• Parquet files: Import structured data files
• DICOM files: Import medical imaging files
• HuggingFace: Download datasets directly from HuggingFace Premium

The app automatically detects classes from folder structure or file organization. Free users can import up to 3 datasets.

OmniClassify supports two main model formats:

• CoreML: .mlmodel or .mlpackage files (Apple's native format)
• ONNX: .onnx files (Open Neural Network Exchange format)

Models are organized by type and task (Classification, Detection, Segmentation). You can filter models by task type in the Models tab. Free users can import up to 2 models.

Yes! Model training is available with a Premium subscription. You can train image classification models using Apple's CreateML framework directly on your device.

All training happens on-device, so your data never leaves your iPhone. Training sessions support data augmentation options and run without iteration limits. Free users cannot access model training.

HuggingFace integration is a Premium feature that allows you to download datasets and models directly from HuggingFace:

1. Get a HuggingFace access token from huggingface.co/settings/tokens
2. Go to Settings in the app and enter your token
3. Browse and download datasets or models directly from HuggingFace
4. Models are automatically converted to ONNX or CoreML format on our server

Important: Not all HuggingFace models are compatible. Compatibility depends on architecture and conversion support. The app will notify you if a model cannot be converted.

Automations & Triggers are Premium features that let OmniClassify take action when detections or classifications meet conditions you define:

• Create detection triggers for specific object classes with minimum confidence thresholds
• Create classification triggers for labels returned during inference
• Choose actions such as sending a local notification, opening a URL, running a Shortcut, or playing haptic feedback
• All automations use App Intents and execute entirely on-device

You can configure automations in Settings > Automations & Triggers, enable or disable them individually, and edit them at any time.

Yes, absolutely! OmniClassify is designed with privacy as a core principle:

• All processing happens on your device—training, inference, and analysis
• Your data never leaves your iPhone
• No cloud uploads or data transmission
• No analytics or tracking
• Works completely offline (except for HuggingFace downloads)
• Complete control over your data

See our Privacy Policy for complete details.

Subscriptions are managed through Apple:

1. Open Settings on your iPhone
2. Tap your Apple ID at the top
3. Tap Subscriptions
4. Find OmniClassify and manage your subscription

You can cancel, change plans, or restore purchases at any time. Refunds are handled by Apple according to their refund policy. You can also tap "Restore Purchases" in the app's Settings.

No. OmniClassify is a general-purpose machine learning toolkit and is NOT a medical device. It is NOT intended for diagnostic, therapeutic, or clinical decision-making purposes.

Results from the app should NOT be used as the sole basis for medical diagnosis or treatment decisions. Always consult qualified healthcare professionals for medical advice.

While the app supports DICOM file import for medical imaging, this is for research and educational purposes only. See our Terms of Service for complete disclaimers.

OmniClassify requires:

• iOS 15.0 or later
• iPhone or iPad
• Sufficient storage space for datasets and models
• For training: More storage space and memory (larger datasets may require more resources)

The app works on all iPhone models that support iOS 15.0+. Training performance may vary based on device capabilities.

Yes! OmniClassify works completely offline except for:

• HuggingFace downloads (requires internet connection)
• Initial app download from App Store

All other features—dataset import, model import, inference, training, camera inference, video processing—work completely offline. Your data never leaves your device.