REI Sat

Inspiration

I got inspired by 'TankerTrackers' where Satelite images are being used to track oil levels and predict the market trends for commodoties.

So I wanted to something with Sattelites, I first thought about undiscovered archeology examples.

What it does

Use satelite images to scan areas and use Pixtral to extract data from the images.

• Homes
• Parking spots
• Railroad
• Schools/Commercial buildings (prediction based on structure of building)

How I built it

I use EarthExplorer for the satelite, but this can be improved upon by using a great API I use Pixtral to get data insights from the images The pixtral generated reporting is stored in a chronological order to then With Mistral AI use all the reports of the area images to aggregate the complete overview and if it's a developing area.

Challenges we ran into

There's a lot of variables that go into the value trends of real estates. So I decided to add the most important ones.

Accomplishments that we're proud of

What we learned

What's next for REI Sat

Improving the Sattelite API integration Improve the factors it checks for, for generating the insights Improving the user workflow and give a better price trend and development comparison Add additional sources of scraping the web for data once an area is targeted "Interesting to look at"

NOTE FOR DEMO: Due to the loading of 100* 5 images and getting the AI insights for all of them is taking time, for the DEMO I picked a fixed location in Cuijk and did all the processing for it. USE image 35,36 on the Cuijk grid if you want to see the specific views they are pre loaded

Built With

• chippai
• mistralai
• pixtral
• sveltekit