CalorieAI Mobile Client

A cross-platform mobile app built with .NET MAUI that estimates nutritional content of meals from photos using AI.

Features

• Photo Analysis — take a photo or pick from gallery to identify dishes and estimate nutrition
• Detailed Breakdown — calories, proteins, fats, and carbs for each detected ingredient
• Custom Notes — add context (e.g., "with soy sauce") to improve accuracy
• History — previous analyses persisted locally via SQLite
• Cross-Platform — built with .NET MAUI for Android and iOS

Screenshots

Tech Stack

Framework	.NET 10 MAUI
Architecture	MVVM (CommunityToolkit.Mvvm)
HTTP	Typed HttpClient via IHttpClientFactory
Error Handling	Result<T> pattern (no thrown exceptions in service layer)
Serialization	System.Text.Json

Getting Started

These instructions walk through running the client on an Android emulator (the quickest path). iOS requires a Mac and is out of scope here.

Clone the repository:

git clone https://github.com/dgluhotorenko/CalorieEstimatorClient.git
cd CalorieEstimatorClient

1. Prerequisites

• .NET 10 SDK
• JetBrains Rider (2024.3+) or Visual Studio 2022 17.12+
• Java 17 JDK (ships with Android Studio, or install Microsoft OpenJDK 17)
• Android SDK + an emulator image — the easiest way to get both is Android Studio (use its SDK Manager and Device Manager), but Rider can also manage the SDK under Settings → Build, Execution, Deployment → Android

Install the MAUI workload (one-time):

dotnet workload install maui-android

Verify:

dotnet workload list

You should see maui-android listed.

2. Start the API first

The client is useless without the backend. Follow the CalorieAI Web API setup to:

1. Configure your Gemini API key in user secrets.

2. Run the API with the http launch profile so it listens on http://localhost:5064:

   cd ../CalorieApi
   dotnet run --launch-profile http

3. Confirm it works by opening http://localhost:5064/swagger in your browser.

Leave the API running in its own terminal for the rest of the setup.

3. Create an Android emulator

If you don't already have an AVD (Android Virtual Device):

• Via Android Studio: Tools → Device Manager → Create Virtual Device. Pick any phone profile and a system image with API level 21 or higher (API 34 recommended). Start the emulator at least once to confirm it boots.
• Via Rider: Settings → Build, Execution, Deployment → Android → AVD Manager.

4. Configure the API endpoint

The default in MauiProgram.cs is already set for the Android emulator:

// MauiProgram.cs
private const string ApiBaseUrl = "http://10.0.2.2:5064";

10.0.2.2 is the special alias the Android emulator uses to reach your host machine's localhost. Keep this value as-is when running in the emulator.

Physical device instead of emulator? 10.0.2.2 won't work. Expose the API through Dev Tunnels (see the API README) and replace ApiBaseUrl with the generated https://*.devtunnels.ms URL.

5. Run from Rider

1. Open CalorieClient.sln.
2. In the run configuration dropdown (top-right toolbar):
   • Project: CalorieClient
   • Target Framework: net10.0-android
   • Device: select your running AVD (start it from the Device Manager if it isn't already up)
3. Hit Run (Shift+F10). First build will restore NuGet packages and may take a few minutes.

The app should deploy to the emulator and open on the main screen. Tap the camera/gallery button, pick a food photo, and you should see a nutrition breakdown within a few seconds.

Running from the CLI (alternative)

cd CalorieClient
dotnet build -t:Run -f net10.0-android

This builds and launches on the first attached/running Android device or emulator.

Troubleshooting

Symptom	Likely cause / fix
Could not reach the server. Make sure the API is running.	API not running, or running under the https profile only. Use dotnet run --launch-profile http.
Build error about missing maui-android workload	Run dotnet workload install maui-android.
Build error: JavaSdkDirectory not found	Install JDK 17 and restart Rider.
Emulator boots but app never appears	Ensure the AVD is fully booted (home screen visible) before starting the run. Cold boots take 30–60s.
HTTP cleartext error in the Android log	usesCleartextTraffic is already enabled in Platforms/Android/AndroidManifest.xml; if you changed it, revert.
Request hangs, then times out	Check that the API is bound to 0.0.0.0 or localhost on port 5064 specifically (the http profile does this). Custom ports need matching ApiBaseUrl.

API Integration

This client communicates with the CalorieAI Web API. Images are sent as multipart/form-data and a structured nutrition report is returned.