Inspiration

Our inspiration was to create an app that works with Real Estate while also adding a blend of Google Map usasge and some level of AI. We brought together all these tools and concepts to craft our perfect idea of an app.

What it does

Our app allows the user to choose a spot in the world using google maps and then generates an image of a house you can expect to find at this country or location using Artificial Intelligence and stable diffusion.

How we built it

We built it by first using Google Maps API and Geocoder API to find user input for the selected area. Once we found out which country or area the user had selected, we inputted the data through a layer of AI's to first find out housing descriptors of that area, then create a prompt using those descriptors and location in ChatGPT, and then finally put in that prompt into a stable diffusion AI to create a perfect house image.

Challenges we ran into

One of the big problems we ran into was getting the APIs to work as Google does not give out free API keys or free access for that matter, unless you want to wait a while. So we had to get around that by setting up our own API account. Another problem was finding datasets for countries/locations and then the descriptors of housing in that location. Without this type of data, we could not build the skeleton of our project.

Accomplishments that we're proud of

Our appliance of the Google Maps API and then displaying it in a working fashion on a front-end website was a big win when we figured it out. Another would be getting down the layering of the 2-3 AIs we used in order to finally generate images at the end.

What we learned

We learned that in order to make a website look good versus maintaining functionality, we require a lot more work and time than what we had these past 36 hours. Another thing we learned was that modern AI like ChatGPT or technology like stable diffusion is incredibly powerful and can make realistic prompts or images on the fly.

What's next for EstateEyes

Hopefully if we can figure out a way to get more datasets on housing descriptors, we can actually get much more precise locations like every city in every country and then really narrow down the types of houses you can find. We would then maybe get more info too, like the typical price of said house in the country, and then maybe link it to real world examples.

Built With

Share this project:

Updates