Astronomer wandering through time and space.
Really just posting about hiking with our dog.
He/Him
| ORCID | https://orcid.org/my-orcid?orcid=0000-0002-8969-5229 |
| Website | https://crawfordsm.github.io/ |
And here is the preprint with more details:
Geospatial foundation models (GeoFMs) promise broad generalisation capacity for Earth observation (EO) tasks, particularly under data-limited conditions. However, their large size poses a barrier to deployment on resource-constrained space hardware. To address this, we present compact variants of a Vision Transformer (ViT)-based GeoFM that preserve downstream task performance while enabling onboard execution. Evaluation across five downstream tasks and validation in two representative flight environments show that model compression and domain adaptation are critical to reducing size and resource demands while maintaining high performance under operational conditions. We further demonstrate reliable on-orbit inference with the IMAGIN-e payload aboard the International Space Station. These results establish a pathway from large GeoFMs to flight-ready, resource-efficient deployments, expanding the feasibility of onboard AI for EO missions.
While producing foundation models is interesting, it is even more interesting when they are put into production. In space!
Science takes all types of contributions and this is a great example of recognizing volunteer and community contributions when the rise to the level of co-authorship.
Since everyone loved my last post on this topic, NASA just dropped a new coloring book about the sun: “Journey Through the Heliosphere: The Sun-Earth system in Color”
Integrity is not only the name of the Orion spacecraft but also a core value at NASA. NASA has policies to label AI generated material and to be transparent with its data. Research integrity and transparency are critical to scientific discovery. From my experience, those working at NASA take this seriously.
So when you see images and if you aren’t sure, check who is sharing it, check the provenance of the images, and check the NASA repositories for the images to verify them.
At the same time, I’ve seen plenty of AI images generated here on social media claiming to be real. With today’s technology, it would not be hard to fake it.
And this raises the important question about trust. While some experiments and research anyone can do, building your own telescope to track Orion is not trivial for anyone. We can’t all do our own research - we have to know who we can trust and we have to build organizations that are trustworthy.
I have gotten my first question on “are the Artemis II images real?” My first thought was: Yes they are real! If they were faked, the toilet wouldn’t have broken.
The Universe is this beautiful and we can do amazing things when we work together.
RE: https://mastodon.social/@crawfordsm/116329229032033234
The Artemis II science team did put together an Open Science and Data Management Plan for their mission.
My very small part in all of it was reviewing it and providing comments. It was very exciting to read about the observations both of the moon and the trip that the astronauts will be doing!
Will it be science and fascinating experiments like Avatar (‘an organ on a chip’)?
