Are you at BMVC tomorrow and interested in VLMs? Come see our poster "Image Recognition with Vision and Language Embeddings of VLMs". 👁️📕 TLDR: We benchmark VLMs on language- and vision-based classification, and propose a simple, training-free vision-language fusion. Link: arxiv.org/pdf/2509.09311

Do I need to say anything?

It is becoming a post-rebuttal tradition to spend the whole weekend reading.

What an opportunity to come visit us in Prague! ❤️ 👀

Nash is almost 5 months and I got used to disappointed faces whenever I show up in the office alone - he mostly comes once a week so I am essentially one big walking disappointment 😂 Maybe I should put his pic on my CV when I start looking for internships 🤔

Have you ever asked yourself how much your favorite vision model knows about image capture parameters (e.g., the amount of JPEG compression, the camera model, etc.)? Furthermore, could these parameters influence its semantic recognition abilities?

Random CVPR story: I went for lunch late, everyone I know had already eaten. Despite being the introvert I am, I thought "it is CVPR, you should meet people," and forced myself to join a guy sitting alone, and the next thing I know, I am on the cover of the computer vision news 😂 (Ralph was great!)

Huge congrats to Illia and Nikita! They started working with us last summer, finished their 1st year of bachelors!! and masters now, and this is their second publication. They worked hard for it alongside full-time courses and didn't get discouraged by the first reject. What can I say, I am proud 🤩

Our paper "Image Recognition with Vision and Language Embeddings of VLMs" has been accepted to BMVC! @bmvcconf.bsky.social 🥳 We compare zero-shot (lang-guided) and vision-only (k-NN) capabilities of recent VLMs (SIgLIP2, RADIO, ..) + propose a simple learning-free fusion, more details coming soon!

Posting cute pics while reading "What is the worst thing your puppy has ever done" reddit threads 😅

1st photo: The day before CVPR 2nd photo: Today Nash is slowly losing the puppy face in exchange for 2.5 extra kgs 😁

Are you at @cvprconference.bsky.social #CVPR2025 ? Come and check out LPOSS. We show how can graph-based label propagation be used to improve weak, patch-level predictions from VLMs for open-vocabulary semantic segmentation. 📅 June 13, 2025, 16:00 – 18:00 CDT 📍 Location: ExHall D, Poster #421

Are you at @cvprconference.bsky.social? Come by our poster! 📅 Sat 14/6, 10:30-12:30 📍 Poster #395, ExHall D

Attending @cvprconference.bsky.social and looking for a PhD or postdoc position in the area of 3d reconstruction (Gaussian splatting, nerfs, scene understanding, etc.)? Find me or drop me an email ;)

⚠️❗Open PhD and Postdoc positions in Prague with Lukas Neumann! ❗⚠️ We rank #5 in computer vision in Europe and Lukas is a great supervisor, so this is a great opportunity! If you are interested, contact him, he will also be at CVPR with his group :)

Join us on June 11, 9am to discuss all things fine-grained! We are looking forward to a series of talks on semantic granularity, covering topics such as machine teaching, interpretability and much more! Room 104 E Schedule & details: sites.google.com/view/fgvc12 @cvprconference.bsky.social #CVPR25

We will be presenting the 🍄 FungiTastic 🍄, a multimodal, highly challenging dataset and benchmark covering many ML problems at @fgvcworkshop.bsky.social CVPR-W on Wednesday! ⏱️ 16:15 📍104 E, Level 1 📸 www.kaggle.com/datasets/pic... 📃 arxiv.org/abs/2408.13632 @cvprconference.bsky.social

IT'S THAT TIME AGAIN!!!! Time to wake up early and go watch birds at the @cvprconference.bsky.social #CVBirdWalk Friday, June 13th, from 7:00 AM to 9:00 AM at the scenic Shelby Bottoms Nature Center & Greenway (1900 Davidson St, Nashville, TN 37206). maps.app.goo.gl/Km8oWJzgVQyo...

What can I say, leaving for CVPR will be hard. It feels like he is growing every day ❤️

Almost tripped over this little guy as I didn’t expect any out during daylight.

Mattia was one of my internship mentors at IBM and I can't recommend him neough!

This week was one of those that make you question whether it would not be better to open a cafe instead of trying to get a phd, but if nothing else comes out of it, at least I can now say I am a phd candidate! #icvv2wacvTrack #bmvc #gaveuponneurips

It is deadline week, you are also defending your thesis proposal/phd state exams in two days, but there is this cutie sleeping on your work chair. What do you do? Yes, she does care if I move the chair. No, she doesn't like any other chair nearly this much.

It is this time of the year in VRG ☕️☀️

Btw Illia is a super talented young guy, just finnishing his 1st year of bachelors, and doesn't know anybody in the field yet, unfortunately nobody from our lab is going to ICLR.. lmk if you are looking to meet new people there and I can connect you! 🙏 👀

Illia will be presenting the poster at ICLR next Friday, come say hi! 📍: Hall 3 + Hall 2B #60, 15:00 🐣 🔍: We made a copy-paste error when creating it, resulting in incorrect labels even in our poster - nobody noticed and I already presented it a few times, can you spot it?

The Visual Recognition Group at CTU in Prague organizes the 49th Pattern Recognition and Computer Vision Colloquium with D. Karatzas, M. Masana, T. Tommasi, P. Mettes @pascalmettes.bsky.social , E. Brachmann @ericbrachmann.bsky.social and V. Stojnic @stojnicv.xyz cmp.felk.cvut.cz/colloquium/#...

Looking for a challenge? Along with FGVC12 at #CVPR2025, multiple fine-grained computer vision competitions have been launched. These well-curated datasets allow you to leverage your AI skills to boost applications with impact! Let's go: sites.google.com/view/fgvc12/... @cvprconference.bsky.social

So I tried Gemini 2.5 for the first time 🥲

We are happy to share LPOSS: Label Propagation Over Patches and Pixels for Open-vocabulary Semantic Segmentation. LPOSS is a training-free method for open-vocabulary semantic segmentation using Vision-Language Models.

🔥🔥🔥 CV Folks, I have some news! We're organizing a 1-day meeting in center Paris on June 6th before CVPR called CVPR@Paris (similar as NeurIPS@Paris) 🥐🍾🥖🍷 Registration is open (it's free) with priority given to authors of accepted papers: cvprinparis.github.io/CVPR2025InPa... Big 🧵👇 with details!

Not sure what to do with all the free time after ICCV deadline?🌴 Are you tired of all the self-driving datasets? Or you just like 🍄? Submit to our FungiCLEF25@CVPR challenge! Just launched, 2 months to go 🏃‍♂️. The task is few-shot recognition of 🍄 observations. More at www.kaggle.com/competitions...

Introducing DaD (arxiv.org/abs/2503.07347), a pretty cool keypoint detector. As this will get pretty long, this will be two threads. The first will go into the RL part, and the second on the emergence and distillation.

Post-ICCV rush mood

I am consistently seeing significant improvements using SigLIP2 (SO) instead of SigLIP across the very diverse datasets I am working with, I very much recommend trying it if you use a CLIP-like model!

Submission deadline is extended to **March 07**. The CMT page is now online and we are looking forward to your submissions on AI systems for expert-tasks and fine-grained analysis. More info at: sites.google.com/view/fgvc12/... #CVPR #CVPR2025 #AI @cvprconference.bsky.social

📢 ‼️ 📢 ‼️ New blog post alert ‼️ 📢 ‼️ 📢 We are finally publishing our blog post: Flaws of ImageNet, Computer Vision's Favorite Dataset. Not only does it have many cute animals 🦦 🐈 🐋 , it also provides an overview of known issues, as well as new insights. klarajanouskova.github.io/ImageNet/

ILIAS: Instance-Level Image retrieval At Scale @gkordo.bsky.social, Vladan Stojnić @annetka.bsky.social Pavel Šuma, Nikolaos-Antonios Ypsilantis @nikos-efth.bsky.social Zakaria Laskar,Jiří Matas, Ondřej Chum, @gtolias.bsky.social tl;dr: SigLIP rules. Lots of ablations arxiv.org/abs/2502.11748 1/

Based on this TMLR got approved for SCOPUS indexing! It may be meaningless for many, but it means my first phd paper might actually get recognized by my uni after all :) docs.google.com/document/d/1... @tmlrorg.bsky.social

When they ask you to review a diploma thesis: 😶 When they mention it will be about gorillas and elephants: 😍

FGVC12 Workshop accepted to CVPR 2025, Nashville! CALL FOR PAPERS: sites.google.com/view/fgvc12/... We discuss domains where expert knowledge is typically required and investigate artificial systems that can efficiently distinguish a large number of very similar visual concepts. #CVPR #CVPR2025 #AI

FGVC12 Workshop is coming to #CVPR 2025 in Nashville! Are you working on fine-grained visual problems? This year we have two peer-reviewed paper tracks: i) 8-page CVPR Workshop proceedings ii) 4-page non-archival extended abstracts CALL FOR PAPERS: sites.google.com/view/fgvc12/...

A late introduction: I am an Associate Professor at CTU in Prague and a computer vision researcher at the Visual Recognition Group (VRG - vrg.fel.cvut.cz). VRG consists of about 40 members and I am leading a small team within VRG. I am made in Greece and exported to France + Czech Republic 🧵⬇️

The CVPR reviewing assignments are so well aligned with what I do I am really looking forward to read all the papers!

Excited to present UDON at NeurIPS '24 tomorrow (Thursday 12/12)! If you are interested in a scalable training method for multi-domain image embeddings, come to poster #1410 in the East Exhibit Hall A-C of the Vancouver Convention Center from 11 am to 2 pm (PST) to discuss!

Here is a 'cat chilling in front of a fireplace' pic to make up for my 'grumpy about datasets' post. 🐈 🐾

What I like to do when considering a new dataset is to train a simple classifier and look at 'the most confident errors'. Recently with NICO: Apart from a class, the images have a context, one of them is 'autumn'. There is also a pumpkin class. Surprise surprise, many autumn images contain pumpkins.

ProbPose: A Probabilistic Approach to 2D Human Pose Estimation Miroslav Purkrabek, Jiri Matas tl;dr: trying to predict whole human pose, not only visible part helps you. It should be done cleverly, though arxiv.org/abs/2412.02254

Segment to Recognize Robustly -- Enhancing Recognition by Image Decomposition @klara-cz.bsky.social, Cristian Gavrus, Jiri Matas tl;dr: sometimes processing background (from GroundingDINO+SAM) and foreground separately helps a lot for hard classification. arxiv.org/abs/2411.15933

My students (1st year msc and 1st year bsc!) submitted to a ML conference for the first time last week, first time any of my students have. They worked so hard for this, in between uni lectures, tests and what not, and I am super excited to share what they have been up to once double blind allows :)