Convergent

Inspiration
Various research papers, Happy City by Charles Montgomery, Orbital Insight What it Does
Uses deep learning to understand urban zone data and tie insights to social factors How we built it
Gathered data, trained multiple models, extracted outputs to visualize Challenges we ran into
Mostly data-related (clouds, resolution, differences between satellites), training time, learning rate, formula for index Accomplishments we’re proud of
Successful AI, identifying changes in urban developments, interpretable results What we learned
Data is key, urban developments are model-able What’s next for Convergent
We’d like to scale our model by geography and scope using the funds we get if we win. Will cover this in greater depth during our presentation.

Built With

• cartodb
• javascript
• numpy
• pandas
• python
• scikit-learn
• tensorflow