🏠 RoomAI: Your Personal Interior Designer Powered by Multimodal AI

This is a submission for the Google AI Studio Multimodal Challenge

What I Built

Our SkillUp30 team @xuanna_chen @bowen007 developed an AI-powered smart home renovation design assistant that leverages multimodal technology to help users quickly achieve professional-grade interior design solutions. This application addresses three major pain points in traditional interior design: high entry barriers, expensive costs, and low decision-making efficiency. Users simply need to upload photos of their rooms and either select preset styles or upload reference style images, and the system instantly generates personalized renovation plans, empowering everyone to become the designer of their own home.

Demo

Project Link: https://smart-home-makeover-ai-317981316224.us-west1.run.app

Core Features Showcase:

• 📸 Upload current room photos with AI automatic space structure recognition
• 🎨 Multiple preset design styles (Modern Minimalist, Scandinavian, New Chinese, Industrial, etc.)
• 🖼️ Support for uploading reference style images for style transfer
• 💡 One-click generation of renovation renderings and design suggestions
• 📋 Detailed material lists and budget estimates

Users can preview different style transformations in real-time and make adjustments based on personal preferences, significantly improving the efficiency and accuracy of renovation decisions.

How I Used Google AI Studio

I fully leveraged Google AI Studio's multimodal capabilities to build this application:

1. Gemini Vision API: For analyzing uploaded room photos, identifying spatial layouts, existing decoration styles, lighting conditions, and other key information
2. Image Generation: Creating high-quality renovation renderings based on user-selected styles and room characteristics
3. Natural Language Processing: Converting visual analysis results into professional design recommendations and material suggestions
4. Style Transfer Technology: Extracting style features from reference images uploaded by users and applying them to target rooms
5. Prompt Engineering Optimization: Carefully crafted prompts to ensure generated renderings are both aesthetically pleasing and practical

Multimodal Features

The multimodal characteristics are the core competitive advantage of this project:

🔄 Image-to-Image Transformation: Converting existing room photos into design renderings of different styles. This visual transformation allows users to intuitively see the post-renovation effects.

📝 Image-to-Text Generation: The system not only generates visual effects but also provides detailed textual explanations, including design concepts, material recommendations, and construction considerations, offering users comprehensive guidance.

🎯 Multimodal Input Fusion: Simultaneously processing users' room photos, style reference images, and text descriptions, integrating multiple input sources to generate solutions that best meet user needs.

💬 Interactive Optimization: Users can further adjust designs through natural language descriptions (e.g., "make the living room brighter," "add more storage space"), and the system understands the intent and updates the renderings accordingly.

These multimodal features significantly enhance the user experience, transforming interior design from a professional service into an intelligent tool accessible to everyone, truly democratizing design
Team Credits: This project was collaboratively developed by the SkillUp30 Team. Our team is dedicated to leveraging AI technology to lower the barriers of professional services, making technology's benefits accessible to more people.
skillup30 team : @bowen007 @xuanna_chen