Build and install the image-segmentation application on an Arm-based Android phone

Connect an Android phone

Enable Developer options and USB debugging on the phone. Connect the phone with a data-capable USB cable, unlock it, and accept the debugging authorization prompt.

Verify the connection:

    

        
        
adb devices -l

    

The output lists the authorized phone and its device serial. If the output reports unauthorized, unlock the phone and accept the debugging prompt. Windows might also need the phone manufacturer’s USB driver. If the phone doesn’t appear, see Run apps on a hardware device .

Prepare the application and model downloader

Clone the example application, then create the Python environment used to download MobileSAM:

    

        
        

mkdir -p "$HOME/image-to-image-android"
cd "$HOME/image-to-image-android"
git clone https://github.com/arm-education/ai-portal-android-app-image-to-image.git
cd ai-portal-android-app-image-to-image

python3 -m venv .hf-venv
.hf-venv/bin/python -m pip install --upgrade pip huggingface_hub
.hf-venv/bin/hf auth login
.hf-venv/bin/hf auth whoami
  

    

    

        
        

$IMAGE_MODEL_WORKSPACE = Join-Path $env:USERPROFILE "image-to-image-android"
New-Item -ItemType Directory -Force -Path $IMAGE_MODEL_WORKSPACE | Out-Null
Set-Location $IMAGE_MODEL_WORKSPACE
git clone https://github.com/arm-education/ai-portal-android-app-image-to-image.git
Set-Location ai-portal-android-app-image-to-image

python -m venv .hf-venv
.\.hf-venv\Scripts\python.exe -m pip install --upgrade pip huggingface_hub
.\.hf-venv\Scripts\hf.exe auth login
.\.hf-venv\Scripts\hf.exe auth whoami
  

    

Sign in with an account that has access to the Arm model repository and use a read token when prompted.

Build and install the application

Build and lint the debug APK, install it, and start the main activity:

    

        
        

chmod +x gradlew
./gradlew :app:assembleDebug :app:lintDebug
adb install -r app/build/outputs/apk/debug/app-debug.apk
adb shell am start -n com.arm.learningpath.imagetoimage/.ui.MainActivity
  

    

    

        
        

.\gradlew.bat :app:assembleDebug :app:lintDebug
adb install -r app/build/outputs/apk/debug/app-debug.apk
adb shell am start -n com.arm.learningpath.imagetoimage/.ui.MainActivity
  

    

The application opens with MobileSAM selected. Its status reports (placeholder bundled) because you haven’t copied the real Hugging Face model into application-private storage yet.

What you’ve accomplished and what’s next

You’ve connected an Arm-based Android phone, prepared the application and model downloader, and installed the debug APK.

Next, you’ll run the MobileSAM model on the phone.