SmartQ by Too Good To Code

Inspiration

To solve Genistat's Solar Challenge, we want to make the process of setting up solar panels easier for both ends of the system - users can quickly request green technology in their housing without a long wait, and experts should be able to set a priority list for more efficient management based on the user's address.

What it does

The app utilizes a specialized algorithm that sorts and ranks future clients' demands upon installing solar panels on their roofs. It is also equipped with a computer vision approach to detect the aerial views of the neighborhood and predict the type of roofs followed by how many solar panels are ideally set up on the residence.

How we built it

We connected to OpenStreetMap's Buildings API and collected the SAT datasets of three regions in Germany: Berlin, Bremen, and Hamburg. We then created a pipeline to process the PBF files and extracted all the necessary information used to train a machine learning model to predict the type of roof and energy efficiency score.

Challenges we ran into

Due to the size of the dataset and the colossal amount of memory it takes on our machines, we had to face multiple waves of utter despair followed by one-hour breaks in a futile attempt to regain what remains of our sanity.

Accomplishments that we're proud of

We successfully set up a working server with Microsoft Azure for the team to code collaboratively. Most importantly, we managed to extract the dataset and computed the needed tilt angle as well as solar efficiency values to answer the questions presented by Genistat. The user interface has been deployed and is available for public viewing under too-good-to-code.streamlit.app.