Mobile Quickstart

Deploy AI models on iOS and Android devices for offline inference and real-time predictions.

What You’ll Accomplish

• Install the SeeMe.ai mobile app
• Download a model to your device
• Make offline predictions
• Capture and label data for training

Why Mobile AI?

Prerequisites

• A SeeMe.ai account (sign up)
• iOS 14+ or Android 8.0+
• A trained model (or use a public model)

Step 1: Install the App

Step 2: Download a Model

After signing in:

1. Tap the Models tab
2. Browse your models or public models
3. Tap the download icon on a model
4. Wait for the model to sync to your device

Step 3: Make Predictions

Camera Mode

1. Tap the Camera tab
2. Select your downloaded model
3. Point at an object
4. See real-time predictions on screen

Gallery Mode

1. Tap the Gallery tab
2. Select your model
3. Choose an image from your photo library
4. View predictions and confidence scores

Step 4: Capture Training Data

Use the mobile app to collect data in the field:

1. Tap Datasets tab
2. Select or create a dataset
3. Tap the + button
4. Capture images with your camera
5. Swipe to assign labels
6. Data syncs to cloud when online

Model Compatibility

Optimizing for Mobile

Export Optimized Models

When training, enable mobile optimization:

from seeme.types import Job, JobType, JobItem, ValueType, Framework, ApplicationType

application_id = client.get_application_id(
    base_framework=Framework.PYTORCH,
    framework=Framework.FASTAI,
    base_framework_version="2.1.0",
    framework_version="2.7.13",
    application=ApplicationType.IMAGE_CLASSIFICATION
)

my_job = Job(
    name="Train mobile model",
    job_type=JobType.TRAINING,
    application_id=application_id,
    dataset_id=dataset.id,
    dataset_version_id=version.id,
    items=[
        JobItem(name="nb_epochs", value="10", value_type=ValueType.INT, label="Number of epochs"),
        JobItem(name="image_size", value="224", value_type=ValueType.INT, label="Image Size"),
    ]
)

my_job = client.create_job(my_job)

Models are automatically converted to Core ML (iOS) and TensorFlow Lite (Android) when auto_convert is enabled on the model.

Model Size Considerations

For mobile, we recommend starting with 224x224.

Offline Workflow

graph LR
    A[Download Model] --> B[Collect Data Offline]
    B --> C[Make Predictions Offline]
    C --> D[Sync When Online]
    D --> E[Retrain Model]
    E --> A

iOS-Specific Features

Shortcuts Integration

Create Siri Shortcuts to run predictions:

1. Open Settings > Shortcuts
2. Create new shortcut
3. Add Run SeeMe Prediction action
4. Configure model and input source

Widget Support

Add a prediction widget to your home screen:

1. Long press on home screen
2. Tap + button
3. Search for “SeeMe”
4. Choose widget size
5. Configure widget model

Android-Specific Features

Share Intent

Predict directly from other apps:

1. Select an image in any app
2. Tap Share
3. Choose SeeMe AI
4. Select your model
5. View predictions

Background Sync

Enable automatic dataset syncing:

1. Open SeeMe app
2. Go to Settings > Sync
3. Enable Background Sync
4. Choose sync frequency

Troubleshooting

“Model failed to download”

• Check your internet connection
• Ensure you have enough storage space
• Try force-closing and reopening the app

“Predictions are slow”

• Use a smaller model (224x224 image size)
• Close other apps
• Ensure device isn’t overheating

“Data not syncing”

• Check internet connection
• Verify you’re signed in
• Check available cloud storage quota

What’s Next?

Video Script Outline