iiinova — where your digital self keeps evolving

Paul Gavrikov @paulgavrikov.bsky.social · 14/07/2026

I am proud to announce that I have joined KAUST (King Abdullah University of Science and Technology) to continue my postdoc journey in the lab of Bernard Ghanem—a very exciting change of scenery as I will be leaving Germany after 30+ years to move to the Red Sea coast in Saudi Arabia 🇸🇦.

120

Paul Gavrikov @paulgavrikov.bsky.social · 07/06/2026

Do Vision-Language Models (VLMs) actually "see" everything in a crowded room? 🔍 Today at #CVPR2026, we are presenting VisualOverload, our work exploring the critical visual perception bottlenecks of VLMs in dense scenes. 📍 Today (Poster Session 6), 5:30 PM - 7:30 PM, Poster 431 (ExHall A)

252

Paul Gavrikov @paulgavrikov.bsky.social · 04/06/2026

Visit our #CVPR2026 poster #179 at 11:50-12:30 to learn about issues and solutions for negation in CLIP. Work led by Fawaz Sammani and Tzoulio Chamiti.

031

Paul Gavrikov @paulgavrikov.bsky.social · 07/04/2026

Meet Slurm Manager: a self-hosted web dashboard for Slurm clusters. Connect via SSH, monitor nodes & jobs in real time, submit scripts, view fairshare quotas — all from your browser. Basically, a handy wrapper over Slurm commands via SSH.

100

Paul Gavrikov @paulgavrikov.bsky.social · 03/04/2026

🚨 Deadline Approaching! 🚨 The archival deadline has officially closed, but there is still time to share your research! The Non-Archival submission deadline for CV4Edu at #CVPR2026 is coming up on April 9th, 2026 (AOE).

100

Paul Gavrikov @paulgavrikov.bsky.social · 26/03/2026

CLIP models excel at image retrieval, but they famously struggle with negation. Do we really need fine-tuning to fix this? 🤔 In our #CVPR2026 MAR workshop paper, "When Negation Is a Geometry Problem in Vision-Language Models," we rethink the problem.

110

Paul Gavrikov @paulgavrikov.bsky.social · 15/03/2026

The CV4Edu @ #CVPR archival deadline ends tomorrow, March 16th AOE! Don't forget to submit your paper on the intersection of computer vision and education. Also, join our workshop in Denver and listen to great keynotes by Marcelo Worsley, @mohitbansal.bsky.social, Jacob Whitehill, and more

101

Paul Gavrikov @paulgavrikov.bsky.social · 12/03/2026

Do you have an interesting paper on the intersection of computer vision and education, but weren't ready to submit it yet? Our extended deadline still allows you to be part of the #CVPR Workshop proceedings! Archival deadline: March 16, 2026 (AoE) Non-archival deadline: May 10, 2026 (AoE)

012

Paul Gavrikov @paulgavrikov.bsky.social · 09/03/2026

🚨 Our submission deadline is approaching soon (March 12, 2026) 🚨 Submit your work to CV4Edu: Computer Vision for Education, our official workshop at #CVPR2026! 📚👁️ 👉 Learn more and submit here: cv4edu.github.io

100

Paul Gavrikov @paulgavrikov.bsky.social · 21/02/2026

Proud to announce that VisualOverload was accepted to #CVPR2026! Overall 2/2 accepted.

020

Paul Gavrikov @paulgavrikov.bsky.social · 27/01/2026

We are proud to announce the first CV4Edu workshop at #CVPR2026 in Denver. We are bringing together researchers in AI in education, computer vision, and human-centered AI, and invite you to submit a full, short, or position paper to our workshop.

123

Paul Gavrikov @paulgavrikov.bsky.social · 14/10/2025

The worst part about leaving academia is that you loose eduroam access :(

100

Paul Gavrikov @paulgavrikov.bsky.social · 01/10/2025

🚨 New paper out! "VisualOverload: Probing Visual Understanding of VLMs in Really Dense Scenes" 👉 arxiv.org/abs/2509.25339 We test 37 VLMs on 2,700+ VQA questions about dense scenes. Findings: even top models fumble badly—<20% on the hardest split and key failure modes in counting, OCR & consistency.

183

Paul Gavrikov @paulgavrikov.bsky.social · 08/09/2025

Is basic image understanding solved in today’s SOTA VLMs? Not quite. We present VisualOverload, a VQA benchmark testing simple vision skills (like counting & OCR) in dense scenes. Even the best model (o3) only scores 19.8% on our hardest split.

2175

Reposted by Paul Gavrikov

Keuper Labs @keuper-labs.bsky.social · 08/07/2025

Congratulations to @paulgavrikov.bsky.social for an excellent PhD defense today!

032

Paul Gavrikov @paulgavrikov.bsky.social · 28/05/2025

Yesterday, I had the great honor of delivering a talk on feature biases in vision models at the VAL Lab at the Indian Institute of Science (IISc). I covered our ICLR 2025 paper and a few older works in the same realm. youtu.be/9efpCs1ltcM

youtu.be

Paul Gavrikov - Feature Biases in Vision Models (Research Talk @ IISc, Bengaluru)

YouTube video by Paul Gavrikov

000

Paul Gavrikov @paulgavrikov.bsky.social · 03/05/2025

What an incredible week at #ICLR 2025! 🌟 I had an amazing time presenting our poster "Can We Talk Models Into Seeing the World Differently?" with @jovitalukasik.bsky.social. Huge thanks to everyone who stopped by — your questions, insights, and conversations made it such a rewarding experience.

130

Paul Gavrikov @paulgavrikov.bsky.social · 24/04/2025

Today at 3pm - poster #328. See you there!

041

Paul Gavrikov @paulgavrikov.bsky.social · 21/04/2025

On Thursday, I'll be presenting our paper "Can We Talk Models Into Seeing the World Differently?" (#328) at ICLR 2025 in Singapore! If you're attending or just around Singapore, l'd love to connect—feel free to reach out! Also, I’m exploring postdoc or industry opportunities–happy to chat!

145

Paul Gavrikov @paulgavrikov.bsky.social · 01/02/2025

We should start boycotting proceedings that do not have a "cite this article" button.

000

Paul Gavrikov @paulgavrikov.bsky.social · 27/01/2025

Proud to announce that our paper "Can We Talk Models Into Seeing the World Differently?" was accepted at #ICLR2025 🇸🇬. This marks my last PhD paper, and we are honored that all 4 reviewers recommended acceptance, placing us in the top 6% of all submissions.

134

Paul Gavrikov @paulgavrikov.bsky.social · 13/01/2025

I analyzed the top 20 most common words in titles of all computer vision (cs.CV) papers on arxiv since 2007. A lot of "learning" papers ...

000

Paul Gavrikov @paulgavrikov.bsky.social · 31/12/2024

10 years ago #ICLR (2014) only had 69 submissions, today nearly 12k. So many good papers: openreview.net/group?id=ICL...

openreview.net

ICLR 2014

Welcome to the OpenReview homepage for ICLR 2014

000

Paul Gavrikov @paulgavrikov.bsky.social · 16/12/2024

Yeah, I am sure my paper on Feature Biases in ImageNet classifiers is a great match for a journal on lifestyle and sustainable goals.

000

Paul Gavrikov @paulgavrikov.bsky.social · 15/12/2024

Today at #NeurIPS2024 Interpretable AI Workshop! Poster sessions are at 10am and 4pm. "How Do Training Methods Influence the Utilization of Vision Models?" arxiv.org/abs/2410.14470 We won’t be there, but Ruta Binkyte is presenting for us!

001

Paul Gavrikov @paulgavrikov.bsky.social · 13/12/2024

Gemini 2.0 Flash is the best free-to-use LLM for writing tasks, blazing fast, and last-but-not-least available in the EU ...

000

Paul Gavrikov @paulgavrikov.bsky.social · 07/12/2024

One of the most interesting aspects of adversarial training to me is not that it increases robustness, but it's ability to revert biases. If a model is strongly biased for X, it will likely completely ignore X at sufficient eps. This is clearly visible in the texture/shape bias!

000

Paul Gavrikov @paulgavrikov.bsky.social · 03/12/2024

The adaption of ChatGPT amongst non-tech users is incredible - I see so many people in public transport using the app.

000

Paul Gavrikov @paulgavrikov.bsky.social · 29/11/2024

Tired of showing the same old same adversarial attack examples? Generate your own with this little Foolbox-based Colab: colab.research.google.com/drive/1WfqEW...

000

Paul Gavrikov @paulgavrikov.bsky.social · 28/11/2024

WILD! Some researchers from have republished ResNet under their own names at some predatory journal. @csprofkgd.bsky.social Predatory: ijircce.com/admin/main/s...

141