Bird Identifier 0.1.5

This update fixes the ARM64 CMake target configuration used to compile the offline TensorFlow Lite runtime.

Identify birds from photographs entirely offline on Ubuntu Touch. Powered by the Google AIY Birds V1 version 3 TensorFlow Lite image classifier. Includes the pretrained model and local label database: approximately 964 bird categories plus background. No account, API key or internet connection is needed.

FEATURES
- Share a photo straight from Gallery to Bird Identifier using Ubuntu Touch Content Hub. Works when the app is already running or launched by sharing.
- Or open a photo from Gallery inside the app.
- See ranked bird species estimates with confidence scores.
- Choose to show 3 or 5 results and set a minimum confidence level.
- Save bird sightings with dates in a private offline journal.
- Light, dark and system appearance modes.
- Top-right hamburger navigation with Settings, Backup / restore and About (version number).
- Export settings and optionally sightings as JSON and restore them on a new phone. Credentials, if configured in future versions, are excluded unless you explicitly select their export; secret-containing JSON is unencrypted.

PRIVACY
Classification happens entirely on the device. No images or recognition data are uploaded. The app requests Content Hub permissions but does not request networking access. Journal entries contain predicted species, confidence and UTC date, not the photo itself.

REQUIREMENTS
Ubuntu Touch 24.04, ARM64. Actual inference speed depends on the device. Classifier predictions are fallible and not definitive species identifications.

LICENSE
Application source code: MIT No Attribution (MIT-0). Bundled Google AIY Birds V1 model: Apache License 2.0. TensorFlow Lite runtime: Apache License 2.0, with separately licensed transitive dependencies. The LICENSE and THIRD-PARTY.md files are included in source.

This application was vibe coded using an AI.
