Inspiration

We were thinking about our annoyances with our rooms and wanted to weaponize AI to crush the bad vibes! So, we decided, Feng Shui For Sure!

What it does

Upload an image or take a quick picture, and Feng Shui For Sure will analyze and rearrange your furniture to maximize your chi! It'll list 3-4 of your top suggested changes.

How we built it

We used Lovable to help build the TypeScript React frontend. After retrieving the image from the user, we called a Python FastAPI backend ran on Google Colab that hosts the Joy AI LLM inference point to return a top down view of the given image. With this top-down image, Google Gemini 2.5 Flash analyzes for a more optimal arrangement in accordance to Feng Shui principles. With these changes, we fed this arrangement into our backend to regenerate an image that incorporates most of these Feng Shui suggestions.

Challenges we ran into

It was difficult to choose the model and also choose how we wanted to host it: we went through many potential tools, such as Vultr, Google Cloud, and Lightning AI, and ultimately settled on Google Colab to host our backend. Additionally, our model still struggles to place certain features of a room correctly due to spacial limitations.

Accomplishments that we're proud of

We all learned a lot about various AI tools, platforms, and processes, and what goes into determining how to integrate AI tools.

What we learned

We learned how to compare and contrast different AI models. We also learned about the specs required to run complex AI inferences.

What's next for Feng Shui For Sure

We would love to include a catalogue with the suggested add-ons. For example, if FSFS recommends adding some plants, it would also provide some links with sources for shopping.

Built With

Share this project:

Updates