1/n Attention, Please! 🚀
Our work “Revisiting Attentive Probing Through the Lens of Efficiency” has been accepted at #ICLR2026.
We introduce Efficient Probing (EP) — a lightweight, multi-query attentive probing method for frozen encoders.
Paper + code at the end 👇
Thrilled to speak at the #mula2026 workshop this morning on Multimodal Large Language Models: Can we make them reliable?
I’m starting at 11:15 in room 111 #cvpr2026 https://t.co/kwAVF49IaO
📍 @circuslegend_ is presenting our work at #CVPR2026 Findings, find the poster with the most 🐈🐾
🗓 Friday, 7:00–8:30
📌 ExHall A; #166
Stop by if curious whether MLLMs make good classifiers or to discuss our "Doomed to Reannotate" ImageNet project - preprint coming soon! 📝
🚨 Call for Papers 🚨
8th Instance-Level Recognition and Generation (ILR+G) Workshop at @eccvconf
📍 Malmö, Sweden
📅 September 8–9, 2026
🌐 https://t.co/nSUFjF3qVb
Submission deadline: June 26, 2026
Notification of acceptance: July 24, 2026
#ECCV2026#ComputerVision
🎉TLDR: We show that retrieving just a few annotated examples can substantially improve open-vocabulary segmentation, bridging much of the gap to fully supervised approaches while preserving open-world generalization.
🏆Team:@Tilemachos77,@stojnvla,Nikos Komodakis,Giorgos Tolias
🚀 Reminder: Our #CVPR2026 highlight ⭐ paper "Retrieve and Segment" will be presented today at the What is Next in Multimodal Foundation Models? Workshop.
🕝 Today, 14:30–16:00
📄 Paper: https://t.co/YPwgKSahph
💻 Code: https://t.co/wWRGYenB5j
Looking forward!
Computer vision means something very different than it did 4 years ago.
I’ll be at CVPR in Denver this week, and the ~4,000 accepted papers tell a story.
Thrilled to have our work listed among the top #CVPR2026 papers! 🎉
RNS is now featured in @skalskip92's curated collection of the most exciting papers from CVPR 2026 👇
🔗 https://t.co/xfawCxXPlW
Big shoutout to @skalskip92 for this incredible effort of tracking 🙌
When I first started thinking about i-CIR dataset, this was one of the images I had in mind.
The Temple of Poseidon in Sounio — not just as “a landmark”, but this particular place, seen again under the iconic condition:
“with a full moon”🌕
🌐Project: https://t.co/9YoLaB8M5F
🚀New dataset: i-CIR — Instance-Level Composed Image Retrieval
💥Given a query image + a text modification, the goal is to retrieve the same particular object after the change — not just any semantically similar object.
📦Dataset: https://t.co/22Zim44LWc
#VLM#AI#MultimodalAI
💥TLDR: Retrieving and leveraging a small set of relevant examples can dramatically improve segmentation quality, narrowing the gap between open-vocabulary and supervised methods.
This video shows the opening ceremony of the 2004 Summer Olympics in Athens, Greece, which marked the return of the Games to their historic birthplace. The ceremony combined theatrical performances, Greek cultural heritage, and a display of sporting unity, as thousands of athletes from 201 delegations made their entrance into the stadium.
The lighting of the Olympic cauldron marked the official start of the Games and celebrated the enduring legacy of the Olympic Games in the modern era.