Smart Wardrobe AI

Inspiration

The "nothing to wear" dilemma is real, despite having a full closet. We wanted to build a tool that removes decision fatigue from our mornings. By combining AI-driven computer vision with real-time environmental data like weather and travel methods, we aimed to create a truly "Smart" wardrobe that thinks for you before you even step outside.

What it does

Smart Wardrobe AI is a digital twin of your closet. It uses the Gemini 3 Flash model to identify and categorize clothing from photos. It doesn't just list your clothes; it acts as a personal stylist. By fetching local weather data and considering your mood and travel plans (like cycling vs. driving), it suggests a complete, visually represented outfit.

How we built it

The app is built using Flutter for a smooth, native Windows experience.

AI: We integrated the Google Generative AI SDK to leverage Gemini 3 Flash for image analysis and reasoning.

Backend & Data: OpenWeather API provides live climate data, while Geolocator handles GPS coordinates.

Persistence: We used shared_preferences to ensure the wardrobe stays saved on the user's machine.

Security: We implemented the envied package to obfuscate API keys within the application binary.

Challenges we ran into

One of the biggest hurdles was the coding of logic behind the application such as how user-defined weather (by dragging along the temperature bar) overwrites the fetched local weather and how we can make the response box show the pictures of AI-picked clothes instead of plain word descriptions. Another challenge was the layouts. In our first sets of tests, the pictures of clothes uploaded were too large and the box containing the response generated by Gemini was not showing the correct format.

Accomplishments that we're proud of

We are particularly proud of the Visual Feedback Loop. Being able to let an AI agent pick clothes for me was like magic. We also successfully implemented a Manual Weather Overwrite, giving users control when GPS isn't available or when they want to plan for a different climate.

What we learned

Building this app was a deep dive into the intersection of Computer Vision and User Experience. We learned how to handle asynchronous data fetching from multiple APIs simultaneously and the importance of obfuscating sensitive keys for public releases. We also discovered that AI is only as good as the context you provide it; adding "Travel Method" changed the quality of suggestions entirely.

What's next for Smart Wardrobe AI

The future of the app involves:

• Multi-day forecasting: suggesting outfits for an entire week based on the 5-day forecast.
• Laundry Tracking: A feature to mark items as "dirty" so they disappear from the recommendation pool.
• Color Palette Analysis: Using AI to suggest outfits that specifically match the user's skin tone or seasonal color palette.

Built With

• envied
• flutter
• geolocator
• google-generative-ai
• image-picker
• openweathermap
• shared-preferences