Furnish It is an AI-powered platform designed to help people furnish their new homes effortlessly. Using data from IKEA and details about your room, our platform recommends personalized furniture options and allows you to visualize them in real-time within a 3D-modeled representation of your space.

Inspiration

When we started the hackathon, we had many ideas to make a project using AI, and best of AI for decision making challenge really resonated with us and we were inspired to help people moving into new residences to design their dream homes.

What it does

The user first uploads photos of the room, which will be turned into a 3D replica. Using the Claude integration, we let the user submit ideas and restrictions. Then Claude will generate search queries, making interior design choices, and a furnished room model will be returned!

How we built it

First, we use an image of a room, coupled with data about the room’s orientation (which is provided by the user via a sleek UI), to create a 3D representation of the room. To simplify the process, the supplied image is used as a HDRI, and the orientation is used to create some shadow-catching planes.

We use a Claude API as a chatbot, and to generate appropriate search queries that will be passed to an IKEA API. We can use this API to also give us 3D furniture models, which will be algorithmically placed in the model room.

Challenges we ran into

We were initially working with an AI model that created a depth map from a single picture for a 3D model of the room. This turned out to be quite challenging due to the perspective distortion causing the 3D model to be highly inaccurate. Thus we ended up asking the user to draw the coordinate axes to map the image to 3D space.

Accomplishments that we're proud of

We’re really proud of how we manage to bring our AI-powered interior design alive through the use of real 3D models from IKEA, rendered with realistic lighting. We are also quite happy with the sleek, user-friendly interface we implemented that makes our application easy-to-use.

What we learned

We learnt a lot about 3D rendering in web browsers, using LLM APIs to solve novel ideas in unique ways that might not seem immediately apparent.

What's next for Furnish It

Moving forward, we’d be interested in creating a full 3D model of the room using a video or multiple pictures. We would also add more integration for ordering the furniture set visualized directly from our platform.

We really enjoyed working on Furnish It, and we hope to continue in the future!

Built With

Share this project:

Updates