Summary of in-progress or completed project.
Doggone It – Dog Breed Identifier is a mobile application that allows users to identify dog breeds using a photo taken with their device camera or selected from the device gallery. The application performs on-device image classification to generate likely breed matches and presents the results in a clear, user-friendly interface.
In addition to identifying breeds, the application allows users to save scans locally, attach notes, and view historical results. For each identified breed, the application retrieves additional breed information from a public dog-breed information API, providing educational details such as temperament, origin, and other characteristics.
The application combines device integration, on-device machine learning inference, local persistence, and external API consumption while maintaining a clean, multiscreen architecture.
1. Dog owner:
2. Dog park visitor:
3. Prospective adopter:
4. Casual learner:
5. Dog lover:
Capture a photo using the device camera
Select an existing photo from the device gallery
Perform on-device breed classification using a TensorFlow Lite model
Display top predicted breeds with confidence values
Save scan results to a local SQLite database
Attach optional notes to saved scans
View scan history in a searchable and sortable list
Mark scans as favorites
Retrieve breed details from an external dog-breed API
Cache retrieved breed information locally for offline viewing
Manage user preferences such as sorting behavior and display options
Generating a locally processed “cartoon-style” portrait when saving a scan
Expanding classification support to include cats
The following information will be stored in persistent device storage so that it remains available across multiple sessions of the application. The app does not rely on a remote server for data persistence; instead, it uses local Android storage (Room database and DataStore preferences).
| Service / Source | Documentation | How the App Uses It | Can the App Function Without It? |
|---|---|---|---|
| TheDogAPI (Breed Information API) | TheDogAPI Documentation | Retrieves detailed information about dog breeds such as temperament, origin, breed group, and other descriptive attributes to display alongside prediction results. | Yes. If the API is unavailable, the app can still perform breed predictions using the local TensorFlow Lite model. Previously retrieved breed information cached in the local database will remain available offline. |
| Dog CEO Dog API (Image / Breed List API) | Dog CEO API Documentation | Provides breed lists and dog images that can be used for browsing breeds or supplementing UI elements such as breed galleries. | Yes. The core functionality of the app (breed prediction) does not depend on this API, and cached or locally stored data can still be displayed offline. |
| TensorFlow Lite (On-Device ML) | TensorFlow Lite Android Guide | Runs the dog breed classification model locally on the device. The app converts captured or uploaded images into a format suitable for the TensorFlow Lite model and receives predicted breed probabilities. | No. Without TensorFlow Lite the application cannot perform breed prediction. |
| Camera (CameraX) | CameraX Developer Guide | Used to capture photos of dogs directly inside the app so the image can be analyzed for breed prediction. | The app cannot capture new dog photos without camera access, but existing photos can still be analyzed, and previously saved scans can still be viewed. |
| Media Gallery Access | Android Shared Media Storage | Allows users to select existing photos of dogs from their device gallery instead of taking a new picture. | The app can still function using the camera, but users will not be able to analyze existing photos. |
| Local Database (SQLite / Room) | Room Persistence Library | Stores scan history, saved predictions, and possibly cached breed information so users can review past scans. | The app could still perform breed predictions, but scan history and stored results would not persist between sessions. |
| Preferences Storage (DataStore) | Android DataStore Guide | Stores small configuration settings such as user preferences, onboarding flags, or display settings. | The app would still function but user settings would not persist between sessions. |
Both TheDogAPI and Dog CEO Dog API provide free access suitable for a student project within their rate limits. TheDogAPI requires an API key, while the Dog CEO API is openly accessible.
Breed information retrieved from these APIs will be cached locally in the application’s Room/SQLite database. Caching reduces repeated network requests and allows previously viewed breed information to remain available when the device is offline.
The application does not require user accounts or authentication. Core functionality such as capturing images and performing breed prediction with the on-device TensorFlow Lite model will continue to function without an internet connection, though new breed information may not be retrieved until connectivity is restored.
Add a feature to create a cartoon style portrait of saved dog photos
Expand classification support to include cats