Alex Stoken

Data Scientist at NASA JSC/JETSII. Views my own.

[Alex Stoken @AlexStoken]

Love timm, but need to do lower-level vision tasks? Meet IMM, a collection of Image Matching Models unified by a simple, easy to use API. You can use any of 30 models (LightGlue, LoFTR, RoMa, SIFT) just by cloning the repo and changing a single parameter! https://t.co/1d65XIMUDo

[Alex Stoken @AlexStoken]

@AJamesMcCarthy Be sure to use the permalink when sharing from this site! https://t.co/QQN9DBUQAK

Links from the address bar are temporary, unfortunately. Permalinks are at the bottom of each search page.

[Alex Stoken @AlexStoken]

@davnords @zhenjun_zhao @gabriberton @isaaccorley_ @Parskatt @ducha_aiki I see it as a nice to have as well, in case someone wants to order kpts to take top N or similar. or take a secondary (higher) threshold w/o rerunning the matcher. It's available in ~30-50% of the matchers, off the top of my head.

[David Nordström @davnords]

We are training a SotA rotationally robust matcher. Currently thinking about what to train for. Comment below your top use-cases for image matching that need more love!

[NASA Earth @NASAEarth]

Our oasis in space 🌎💙

Seeing Earth from this unique perspective reminds us how special our home planet is. Download the 2026 Earth Day poster for a daily dose of inspiration: https://t.co/4Ptd53EoFY

[Alex Stoken @AlexStoken]

Let’s see if it can unseat RoMa atop the model download tracker 👀 https://t.co/jatI181bWT

[Alex Stoken @AlexStoken]

When @davnords @Parskatt release a new SoTA matcher, we try to get it on pypi as soon as possible! Thanks to @davnords for the PR.

Currently the easiest way to use LoMa-R, the rotation invariant version 🔄

[Gabriele Berton @gabriberton]

VisMatch 1.2.1 is out!

And it includes the state-of-the-art matcher LoMa!

Congrats and thanks to @AlexStoken @davnords @Parskatt

[Alex Stoken @AlexStoken]

@gabriberton @davnords @Parskatt @zhenjun_zhao @isaaccorley_ Their is also a bunch of orthomosaics from Landsat/sentinel with really reliable per px location, which can make for easy GT. Good for seasonal variation, but will need the ISS photos or data aug for perspective change 😉 but I ofc course thing this is a great idea

[Alex Stoken @AlexStoken]

@Parskatt @zhenjun_zhao @isaaccorley_ @gabriberton @davnords excited to see if it outperforms the cross modal trained models… we’ll give it a shot!

[Alex Stoken @AlexStoken]

@Parskatt I have a feeling we can find a use for this somewhere! Celestial examples are really impressive. Nice work!!

[Alex Stoken @AlexStoken]

@gabriberton @Ichsan2895 Ya I have some spaghetti code for this that needs organizing, but would also love to see a PR!

[Alex Stoken @AlexStoken]

@Rafael_L_Spring @gabriberton Reports of matching’s demise were greatly exaggerated 😉

[Alex Stoken @AlexStoken]

Vismatch (formerly image-matching-models) is out on pypi!

One line install, one line matching, one simple API.

Very excited about this milestone after some awesome recent updates in collab with @gabriberton @ariG23498 !

[Gabriele Berton @gabriberton]

VisMatch is on pypi!

VisMatch is a wrapper for image matching models, like LightGlue, RoMa-v2, MASt3R, LoFTR, and 50+ more!

It's literally as simple as:

pip install vismatch
vismatch-match --inputs img0 img1 --matcher choose_any

To run image matching on any 2 images [1/4]

[Gabriele Berton @gabriberton]

Sorry about it but AstroLoc is just too good

Blurry aerial footage localized in literally a second across Earth

[Alex Stoken @AlexStoken]

A good detector can make or break your spacecraft localization pipeline, and this is a great opportunity to try your hand at designing a SOTA, generalizable detector!

[Alex Stoken @AlexStoken]

Here’s your chance to contribute to the next generation of vision-based navigation in space!

[topcoder @Topcoder]

🚀 The NASA $55,000 Crater Detection Challenge is LIVE!

Can your algorithm spot craters on the Moon?

Join the Crater Detection MarathonMatch and help NASA improve optical navigation systems for future lunar missions.

Link in the comments.

#NASA #AI #DataScience

[Alex Stoken @AlexStoken]

In cis-lunar space, craters are descriptive, reliable landmarks for pose-estimation. They have distinct shapes and can be matched to a compact catalog of known craters.

[Alex Stoken @AlexStoken]

@isaaccorley_ @gabriberton @ariG23498 @motokimura1 Oh that’s really good to hear, thanks!

[Alex Stoken @AlexStoken]

@gabriberton @ariG23498 @isaaccorley_ @motokimura1 Definitly interested. I have a few more questions, but I'll move discussion to the issue @ariG23498 opened and respond more thoroughly there.

[Alex Stoken @AlexStoken]

and of course, ISS in Real time features plenty of astronaut photography of Earth! Check it out:

https://t.co/mF9e7JacwC

[Alex Stoken @AlexStoken]

Just in time for the 25th anniversary of the International Space Station, “ISS in Real Time” lets you peek into the astronaut experience on humanity’s premier orbiting outpost.

Another incredible project from @BenFeist and Dave Charney!

[Ben Feist @BenFeist]

Announcing ISS in Real Time, a new multimedia project where you can play back every day of the past 25 years aboard the International Space Station.
https://t.co/hoqV5DEH8z