Segment Solar Panels in Aerial Imagery for an Energy Audit Startup
Overview
What this challenge is about.
You receive 600 labelled 1024x1024 orthophoto tiles (panel masks) and 1,000 unlabeled tiles. Train a segmentation model (U-Net or DeepLabV3+ baseline), validate at 0.85 IoU on a held-out 100-tile set, and build a Streamlit-based correction tool that lets an annotator brush-edit the model's masks. Report per-rooftop time-savings on a small 5-rooftop pilot (model + human vs. human-only). Deliver: trained model, the Streamlit tool, an evaluation report, and a one-page rollout plan.
The Brief
What you'll do, and what you'll demonstrate.
Triple annotator throughput on rooftop solar audits by pairing a segmentation model with a fast correction tool.
Earning criteria — what you'll demonstrate
- Train a semantic-segmentation model on aerial imagery
- Build a thin human-in-the-loop correction interface
- Evaluate segmentation with IoU on a clean held-out set
- Measure time-savings of a model-plus-human workflow against human-only
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Computer Vision Engineer
Shipping a segmentation model plus a correction tool on real aerial imagery is core CV-engineer work at climate, mapping, and insurance-tech startups.
This challenge sharpens
- semantic-segmentation
- u-net
- aerial-imagery
Machine Learning Engineer
Pairing a model with a human-in-the-loop tool plus an honest time-savings measurement mirrors what MLEs ship for ops teams.
This challenge sharpens
- human-in-the-loop
- evaluation
- semantic-segmentation
AI Engineer
Standing up the Streamlit correction tool plus a rollout plan is exactly the glue-engineering AI engineers do at vertical-AI startups.
This challenge sharpens
- streamlit
- human-in-the-loop
- evaluation