android

Selfie2Anime Android App

This is an Android sample app with the Selfie2Anime TensorFlow Lite model.

Please read the detailed instructions on how to create the application from scratch in this blog post on Medium: Part 3 Android app.

TensorFlow Lite Model

The TensorFlow Lite Model used in the Android app is located in this project repo here.

|--ml  
|  |--add-meta-data-CLI  
|  |  |--model_with_metadata  
|  |  |  |-- selfie2anime.tflite  

Alternatively you could use the colab to add metadata to the .tflite file.

Requirements

• Android Studio Preview Beta version - download here.
• Android device in developer ode with USB debugging enabled
• USB cable to connect an Android device to computer

Build and run

• Clone the project repo:
  git clone https://github.com/margaretmz/selfie2anime-with-tflite.git
• Open the Android code android/selfie2anime in Android Studio.
• Connect your Android device to computer then click on "Run -> Run 'app'.
• Once the app is launched on device, grant camera permission.
• Take a selfie and wait for the TensorFlow Lite model to process the selfie.
• You will then see a screen with both the selfie and anime image.

Download apk

If you are not familar with building the app in Android and would like to just try out the app, please download the apk file from the android/app/release/ folder.

Note: you will need to grant permission to install unknown apps from your Android device security settings.