An Android app to calculate the value of coins in an image using a TensorFlow Lite image recognition model.
The app works by first capturing an image of coins (all placed to show the reverse (or the "tails") of the coin). Then, the app displays an image of found coins for the user to select or deselect coins. Once the users are happy with the selection, they can initiate the coin recognition, see the calculated sum of values and update the wrongly recognized coins to their correct values.
The app uses intents to allow the user to either capture an image of the coins or select one from the gallery. Then, it uses OpenCV and it's Hough Circles transformation method to find circles in the image (which hopefully correspond to coins). Later, these circles are used in a TensorFlow Lite model, generated using Firebase, to predict the values of each of the coins.
- TensorFlow Lite image recognition
- Google ML-Kit
- OpenCV image processing
- Hilt dependency injection
- Model-View-ViewModel architecture
- Android Jetpack
- LiveData objects
- Android RecyclerViews
- Android Intents
- Android custom AlertDialog
