Panel Efficiency using 3D Mesh Data

Inspiration

Distribute solar panels in a data-driven way.

What it does

Predict best location for solar panels in terms of panel efficiency

How we built it

• Reverse Engineered Google Earth to obtain 3D Mesh data
• Use this to calculate roof tilt to get better panel efficiency estimates
• Train AI to predict this from satellite images to reduce cost

Challenges we ran into

• Rate limits
• No available data set
• Short time span

Accomplishments that we're proud of

• First data set for roof data for buildings created in a short time

What we learned

• Reverse engineering
• Working with meshes
• Working with map rendering
• Working with geo data

What's next for Panel Efficiency using 3D Mesh Data

• Improve AI model
• Merge OSM data with Mesh data
• Expand data set to entirety of Germany
• Actively predict buildings instead of allowing exploration

Built With

• pymesh
• python
• pytorch
• streamlit