Video wird geladen...
Video konnte nicht geladen werden
supervision just hit 40,000 GitHub stars! it now powers over 6.5k open-source computer vision projects, including all my demos like basketball AI link:
890,558 Aufrufe • vor 3 Monaten •via X (Twitter)
67 Kommentare

person of interest, one of my favorite TV series

@averageandy would be cool if you could use this to predict in the first few minutes of a game if a player is playing poorly or well compared to their baseline, then predict the outcome based on all players how they’re playing that day

@averageandy that is not a crazy idea

I know this is super usefull, but its also so cool to watch 😀 Congrats on that huge amount of stars 💪

thank you! we have a lot of new ideas around pose estimation and VLMs.

I'll be watching carefully 😀👀

It is a masterpiece in cv

we have so many new ideas lately; I feel we will get 50k soon

Github stars like a waterfall

i detected 41.1k stars

this is coool! what are you using / building?

Wow that’s amazing

thank yoooou!

curious how you're counting the 6.5k projects is that github dependency graph data or are you tracking it some other way?

yup, dependency graph. there is no other way to do it as far as I know.

@PINTO03091 i keep on seeing projects similar to super pinto like this. Im not very familiar with this area of work, is this the same thing that you are doing or do you take a different approach. it seems like your models scan more data in these videos.

@PINTO03091 I need to try integrationg @PINTO03091's models into supervision

for things like basketball is it sam3.1? live?

It is SAM3.1 combo with other model

Its nice but ive read on their benchmarks that they could do around 20fps for tracking 20 at the same time so it cant do live tracking of everything for live high res feeds no?

Man this looks so cool, makes me want to play around with the code immediately even though I’ve never worked with computer vision

you use coding agents?

Yup, thinking of a little proof of concept to build

I need to make this a lot more accessible to non-coders

I think my main issue will be finding source media. Any tips?

Are there any demoes of using it for test automation of web pages and executables?

I have not played with that usecase :/ sorry

going to give it a shot

Banger

40k stars and powering 6.5k projects is insane, computer vision finally has its go to open source library 🔥

Anybody know the actual repo for the basketball app?

Notebook sits here for now: Search: basketball. I need to move it at some point.

bro this is too sick thank you so much

@Scobleizer This is fantastic

I LOVE YOU!!

@Scobleizer That’s amazing !

This is dope

This is awesome!

thank you!

can this detect different faces?

nope, I'm not working on things like this

☹️

Are you the author?

I started it and I'm still top 1 contributor, but there are 150+ contributors.

I've got a stress test for your code. Can it detect the exact image frame in each of these scans of an old educational filmstrip? And if it can, I'd be curious to see if it can do faded ones, too:

I'm not a coder in any sense of the word so I feel like I would not be able to positively confirm whether or not it could if I tried. But if you or someone else could figure this out you would be helping a very underfunded and understaffed historical preservation project.

Incredibly impressive

thank you!

amazing job!

40k stars on a CV library is a real milestone. Curious how the community is using it outside of demos, especially in robotics pipelines where real-time constraints hit differently than batch inference.

Thank you for building it open source! After seeing this post went to try it and it is super cool

thank you, let me know what are you building

Cool

Does it do football?

it help you to do it; I made this football demo few years ago: it's a bit outdated. it could be a lot better now.

Real football 🤣 American Football

lol

@MaxScore

@Scobleizer How does it perform with industrial error detection (scratches etc)

@Scobleizer @roboflow (the company I work for) is computer vision provided for biiig manufacturing companies

Great to see

@0xbianc8

for the basketball player, how did you make it distinguish the player’s name? it’s awesome

long story hahaha; I have whole tutorial covering this project

@izzatraihan can this be applied to millions of people walking down the streets?

@izzatraihan not sure what's the question

40k stars is actually insane. huge congrats on that, man 🫡
