Kosta Derpanis

#CS Associate Prof @YorkUniversity #ComputerVision Researcher, @VectorInst Affiliate @UofT status only, @ELLISforEurope Member, X-social media chair 4all vision

[Kosta Derpanis @CSProfKGD]

I’ve launched 2 new course websites for my Introduction to #ComputerVision and #DeepLearning in Computer Vision courses.

🆕 Gaussian Splatting
🆕 Flow Matching
🆕 Sparse Autoencoders
🆕 Lecture Notes (work in progress)

The videos don’t include voiceovers, but that’s coming in a future revision 🤞

Share your feedback in the thread.

[Kosta Derpanis @CSProfKGD]

Nobody's going to know 😉 (includes audio)

[Kosta Derpanis @CSProfKGD]

@beckysoltanian 🤔

[Kosta Derpanis @CSProfKGD]

Nobody's going to know 😉 (includes audio)

[Kosta Derpanis @CSProfKGD]

The tale of an AI slop paper, from creation to submission.

[Kosta Derpanis @CSProfKGD]

@sitzikbs @CVPR I’m not part of the organizing committee, but let’s ask a #CVPR2027 PC @humphrey_shi

[Mark Gurman @markgurman]

The Amazon Ring in-home drone security camera is finally actually shipping - January for $250. https://t.co/01lUd9vk6z

[Kosta Derpanis @CSProfKGD]

My part-time office for the next week ☕️

Writing and editing #CVPR2027 papers.

[Stavros Diolatzis @StavrosDiol]

If you are attending WACV 2027 have a look at our latest paper on combining feature grids with 2D Gaussians! Grids are good at representing areas of images with uniform frequency content so we use 2D Gaussians to fit the frequency maps of images!

https://t.co/yS6xKa8UsJ

[Jonathan Stephens @jonstephens85]

Aw snap!!! Someone on Reddit put 3DGS on an old iPod Nano. This blows my mind. Just a couple years ago this required high end GPU.

[Kosta Derpanis @CSProfKGD]

@DominiqueCAPaul Surprised? This is reality for many students. Some face long waits, others can’t even apply.

[Kosta Derpanis @CSProfKGD]

@Michael_J_Black robust hierarchical flow versus phase-based flow

[Anthony Bonato @Anthony_Bonato]

Student in 2020: Yo Prof, I don't get #6

Student in 2026: Dear Esteemed Professor, I hope this email finds you well. I am reaching out to respectfully express some conceptual challenges regarding question 6. Your guidance would be invaluable to my academic journey

[Peter Richtarik @peter_richtarik]

yup, happens

[Michael Black @Michael_J_Black]

One of the most fun and productive periods of my career involved what I'd call "cooperative competition".

I was a postdoc in Toronto, working with Allan Jepson. He and I both had different methods for computing optical flow from video.

It was a like a dance. I would improve my method and run into his office with the new results. He'd then improve his method and run into my office with his results.

We were learning from each other and the competition drove each of us to be better. It was exciting and joyful.

If you find yourself in such a situation, enjoy it. I think it's rare. Put aside ego and have fun.

[David Lindell @DaveLindell]

New Gaussian splatting representation based on polyline primitives! The primitives are optimized directly from images and can form long, connected chains. Perfect for reconstructing strand structures (spaghetti, hair, etc.) and it matches 3DGS performance on general scenes.

[Sara Beery @sarameghanbeery]

🌍🔬 I’m recruiting PhD students for the BeeryLab at MIT (starting Fall 2027)!

We build computer vision and machine learning methods for biodiversity monitoring at a global scale. Learn more about our research here: https://t.co/GzA3vATLil

[Jahir Sheikh @jahirsheikh8]

Bro got rejected by HR and the entire engineering department 💀

[Linus ✦ Ekenstam 🔜 Adobe MAX Miami 2026 @LinusEkenstam]

Adobe is back, officially releasing MCP support for its entire creative cloud stack.

they are back on their toes, feels exciting to see this be released in real-time in Miami.

Adobe becoming headless!?

[Goodfire @GoodfireAI]

In Neural Networks are Already Clean, @nickcammarata argued that neurons were far more legible than previously thought – and that this was great news for interpretability and safety.

Now, we’re releasing the first technical companion to Nick’s essay. It shares our findings from studying thousands of neurons using our novel automated interpretability pipeline, and outlines why we now think of neurons by their computational function, instead of whether or not they are polysemantic. Here’s what we found: 🧵

[Andrea Tagliasacchi @taiyasaki]

📢📢📢 Introducing Spaghetti Splatting

3DGS is a convolution of a Dirac with a Gaussian Kernel. We provide a strict generalization of 3DGS, extending its Dirac mass to lines and curvilinear structures

project page: https://t.co/oGPJAuBE4O
arXiv: https://t.co/2Ql1Edi9JJ

🔊 on!