Locamage

Understanding the Earth

[Locamage @Locamage]

Every example comes from a public benchmark: FLAIR2, SpaceNet2, m-Chesapeake and HLS Burn Scars.

See them all, with a slider between imagery and map: https://t.co/ZYm27RmQRv

[Locamage @Locamage]

Day 2 of Finch 1: what it sees.

Land cover in France, buildings in Khartoum, roads and roofs around the Chesapeake Bay, and a wildfire scar in the US, from 20 cm aerial photos to 30 m satellite imagery.

Every map comes from one frozen model and a single linear layer.

[Locamage @Locamage]

Meet Finch 1, our first step toward one model that works with any sensor, at any resolution, anywhere on Earth.

Trained on imagery from 50+ sensors across satellites, aircraft and drones.

[Locamage @Locamage]

@speedrun Thank you Speedrun Alpha for betting on us!

[Locamage @Locamage]

Meet Finch 1, our first step toward one model that works with any sensor, at any resolution, anywhere on Earth.

Trained on imagery from 50+ sensors across satellites, aircraft and drones.

[Locamage @Locamage]

Full results and how we compared: https://t.co/qGf6DRTFUb
Request early access: https://t.co/KSlpYV7zmx

[Locamage @Locamage]

If you are a researcher and want to work with us, email [email protected].

Thanks to a16z Speedrun Alpha (@speedrun) and Google's TPU Research Cloud and TPU Builder Program for supporting this work.

[Locamage @Locamage]

And it keeps getting better as we scale.

Full results and how we compared: https://t.co/qGf6DRTFUb

[Locamage @Locamage]

12 hours

[Locamage @Locamage]

Finch