Find your images quickly! mytag aims to be a simple auto-tagging image gallery.
Powered by a Tensorflow model with on-device classification. Built with Expo+React Native.
Right now, the first prototype achieves:
- On start, asks user for media library permissions. On the background, the app initializes the pre-trained model (COCO-SSD).
- When given, the app grabs all available images on the device and classifies one by one, caching results in local storage.
- After each new result, the app shows the currently selected category and images associated with it (photos with a
personin it, for example). - From the dropdown, the user can choose different categories and view the corresponding images.
- If the app is open later, the "analysis" will be much much faster, as results were cached locally.
APK available: find these features and more in Releases
You can either use the Docker container, or install Expo.
- Install Docker Engine and Docker Compose.
- Clone and start the container:
git clone git@github.com:cryptoboid/mytag.git
cd mytag
docker-compose run expo- Duplicate
.env.templateand rename to.env. If you will build/publish on Expo, complete your username and password, otherwise they are not necessary. - Once inside, start the expo server (with hot reloading):
expo start - Download the Expo Go app in your phone and test it!
- Install Expo CLI:
https://expo.io/tools#cli - Clone repo:
git clone git@github.com:cryptoboid/mytag.git
cd mytag- Install npm packages:
npm install - Duplicate
.env.templateand rename to.env. If you will build/publish on Expo, complete your username and password, otherwise they are not necessary. - Start the expo server (with hot reloading):
expo start - Download the Expo Go app in your phone and test it!
You can see our to-do list in this project. Feel free to work on any of these issues!
MIT