iiinova — where your digital self keeps evolving

Reposted by Nikos Efthymiadis

Bill Psomas @billpsomas.bsky.social · 06/01/2026

🚀New task: Instance-level Image+Text→Image Retrieval 🔎Given a query image + an edit (“during night”), retrieve the same specific instance after the change — not just any similar object. 🛢New dataset on HF: i-CIR huggingface.co/datasets/bil... 🔥Download, run, and share results!

0125

Reposted by Nikos Efthymiadis

Bill Psomas @billpsomas.bsky.social · 06/11/2025

⬇️ Grab i-CIR, run your method, tell us how it handles instance-level composed image retrieval. 📄 arxiv.org/abs/2510.25387 🧪 github.com/billpsomas/i... George Retsinas, @nikos-efth.bsky.social, Panagiotis Filntisis, Yannis Avrithis, Petros Maragos, Ondrej Chum, @gtolias.bsky.social.

arxiv.org

Instance-Level Composed Image Retrieval

The progress of composed image retrieval (CIR), a popular research direction in image retrieval, where a combined visual and textual query is used, is held back by the absence of high-quality training...

031

Reposted by Nikos Efthymiadis

Dmytro Mishkin @ducha-aiki.bsky.social · 03/11/2025

Instance-Level Composed Image Retrieval @billpsomas.bsky.social George Retsinas @nikos-efth.bsky.social Panagiotis Filntisis,Yannis Avrithis, Petros Maragos, Ondrej Chum, @gtolias.bsky.social tl;dr: condition-based retrieval (+dataset) - old photo/sunset/night/aerial/model arxiv.org/abs/2510.25387

1154

Reposted by Nikos Efthymiadis

Giorgos Tolias @gtolias.bsky.social · 01/09/2025

The Colloquium in Pattern Recognition and Computer Vision of the Visual Recognition Group at CTU in Prague has a long tradition dating back to 1998. The list of all speakers is available docs.google.com/spreadsheets.... Enjoy! The 50th edition is coming soon cmp.felk.cvut.cz/colloquium/

docs.google.com

Pattern Recognition and Computer Vision Colloquium - past speakers

1178

Reposted by Nikos Efthymiadis

Giorgos Kordopatis-Zilos @gkordo.bsky.social · 22/06/2025

🚨 Deadline Extension Instance-Level Recognition and Generation (ILR+G) Workshop at ICCV2025 @iccv.bsky.social 📅 new deadline: June 26, 2025 (23:59 AoE) 📄 paper submission: cmt3.research.microsoft.com/ILRnG2025 🌐 ILR+G website: ilr-workshop.github.io/ICCVW2025/ #ICCV2025 #ComputerVision #AI

11210

Reposted by Nikos Efthymiadis

Giorgos Kordopatis-Zilos @gkordo.bsky.social · 28/03/2025

Exciting News! Shoutout to @hf.co 🤗 for hosting ILIAS in its full scale! You can now easily download it via: huggingface.co/datasets/vrg... This marks a milestone in i2i and t2i retrieval, paving the way for significant advancements in the field. Huge thanks to @hf.co team for their support!

huggingface.co

vrg-prague/ilias · Datasets at Hugging Face

We’re on a journey to advance and democratize artificial intelligence through open source and open science.

1298

Reposted by Nikos Efthymiadis

Giorgos Tolias @gtolias.bsky.social · 03/03/2025

The WACV'25 work of @nikos-efth.bsky.social is on WACV daily - computer vision news.

082

Reposted by Nikos Efthymiadis

Giorgos Kordopatis-Zilos @gkordo.bsky.social · 27/02/2025

ILIAS is a large-scale test dataset for evaluation on Instance-Level Image retrieval At Scale. It is designed to support future research in image-to-image and text-to-image retrieval for particular objects and serves as a benchmark for evaluating foundation models and retrieval techniques.

23915

Reposted by Nikos Efthymiadis

Dmytro Mishkin @ducha-aiki.bsky.social · 24/02/2025

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/

1259

Reposted by Nikos Efthymiadis

Giorgos Tolias @gtolias.bsky.social · 12/02/2025

For PhD and MSc students interested in a research visit to Prague/VRG in 2025: we're open to hosting short-term collaborations or internships on a range of computer vision topics. If this sounds exciting, reach out by e-mail! We'd love to discuss potential projects. Some examples 🧵 #Internship #CV

12722

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

7/ Thanks to my supervisors and co-authors @gtolias.bsky.social and Ondrej Chum for making this possible!

110

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

6/ Check out our WACV resources: 🎥 Video presentation: youtu.be/ZPw3CUioMMg 🖼️ Poster: cmp.felk.cvut.cz/~efthynik/po... 🌐 Project homepage: cmp.felk.cvut.cz/~efthynik/cr...

youtu.be

[WACV 2025 Oral] Crafting Distribution Shifts for Validation and Training in Single-Source DG

YouTube video by Nikos Efthymiadis

100

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

5/ Plus, we introduce a state-of-the-art single-source-domain-generalization method, setting new records on four popular generalization benchmarks.

100

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

4/ Our work offers a solution: 🌟 Augmenting the validation set! We achieve correlations of 0.74 (PACS) and 0.83 (Mini-DomainNet)—enabling confident model selection without access to the target domain!

100

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

3/ But what if your validation-to-test correlation looked random (left scatter plots)? 😬

100

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

2/ Here’s the setup: Imagine building a model to classify photos 📸, but your goal is to generalize to sketches ✏️ or paintings 🖌️. You only have labeled photos. The target domains are inaccessible due to cost or legal rights. You validate models on the source domain and hope for the best.

100

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

🧵 1/ Excited to share that our WACV 2025 paper, "Crafting Distribution Shifts for Validation and Training in Single Source Domain Generalization," was also selected for an Oral Presentation! 🎤

160

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

5/ A huge thanks to my amazing co-authors: @billpsomas.bsky.social, Zakaria Laskar, Konstantinos Karantzalos, Yannis Avrithis, Ondrej Chum, and @gtolias.bsky.social .

110

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

4/ check out our resources: 🎥 Video presentation: youtu.be/l57-rGT6zEs 🖼️ Poster: cmp.felk.cvut.cz/~efthynik/po... 🌐 Project homepage: cmp.felk.cvut.cz/~efthynik/fr...

youtu.be

[WACV 2025 Oral] Composed Image Retrieval for Free Domain Conversion

YouTube video by Nikos Efthymiadis

110

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

3/ 🖼️ We expand ImageNet-R to include more domains as image queries and introduce three new benchmarks, creating a more stable research base. 💡 We propose FreeDom: a novel, training-free approach, using the concept of memory-based textual inversion. 🚀

110

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

2/ The challenge: Domain conversion 🎨 is an underexplored task. Most efforts have focused on photo-to-rest conversions, like ImageNet-R. But our work expands the scope in two major ways:

110

Nikos Efthymiadis @nikos-efth.bsky.social · 28/01/2025

🧵 1/ Big news! Our WACV 2025 paper, "Composed Image Retrieval for Training-Free Domain Conversion", has been selected for an Oral Presentation! 🎤✨

141

Reposted by Nikos Efthymiadis

Giorgos Tolias @gtolias.bsky.social · 07/01/2025

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 🧵⬇️

vrg.fel.cvut.cz

VRG - Visual Recognition Group

1439

Reposted by Nikos Efthymiadis

Nikolaos-Antonios Ypsilantis @nikosips.bsky.social · 12/12/2024

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!

0126

Nikos Efthymiadis @nikos-efth.bsky.social · 09/12/2024

4/4 Also, we introduce a single-source-domain-generalization method, which is the new state-of-the-art on 4 benchmarks. Our paper on ArXiv: www.arxiv.org/abs/2409.19774 Our code on GitHib: github.com/NikosEfth/cr... Shout-out to my supervisors and co-authors @gtolias.bsky.social and Ondrej Chum

arxiv.org

Crafting Distribution Shifts for Validation and Training in Single Source Domain Generalization

Single-source domain generalization attempts to learn a model on a source domain and deploy it to unseen target domains. Limiting access only to source domain data imposes two key challenges - how to ...

010

Nikos Efthymiadis @nikos-efth.bsky.social · 09/12/2024

3/n It offers a practical alternative to this problem! With domain-agnostic augmentations, we achieve correlations of 0.74 (PACS) and 0.83 (Mini-DomainNet)—letting you make confident model choices even without target domain access.

100

Nikos Efthymiadis @nikos-efth.bsky.social · 09/12/2024

2/n But what if the validation-to-test correlation looked like the scatterplots on the left? 😬 They are random at best, aren't they? We are happy to present our second WACV 2025 paper named: "Crafting Distribution Shifts for Validation and Training in Single Source Domain Generalization".

100

Nikos Efthymiadis @nikos-efth.bsky.social · 09/12/2024

1/n Bear with me for a moment: Imagine you train a model to classify photos, but you need it to generalize to paintings 🎨. You only have data from photos, and the target domain is inaccessible due to cost or legal rights. You train many models and you validate on the source domain.

182

Nikos Efthymiadis @nikos-efth.bsky.social · 06/12/2024

Thank you for the kind words, Philip! The closest dataset to DeepFashion’s Consumer-to-Shop in our paper is Fashion-IQ, and detailed performance results are included in the appendix. Also, FreeDom is designed for domain conversion, but DeepFashion is on our radar for our next, more general work.

010

Reposted by Nikos Efthymiadis

Bill Psomas @billpsomas.bsky.social · 05/12/2024

🚀 New paper alert: FREEDOM is here! Check out “Composed Image Retrieval for Training-FREE DOMain Conversion,” our training-free method for domain conversion with VLMs. 🎯 📜 WACV 2025 💡 Retrieve images using image+text queries! 📖 arxiv.org/abs/2412.03297 🔗 github.com/NikosEfth/fr... #WACV2025 #AI

arxiv.org

Composed Image Retrieval for Training-Free Domain Conversion

This work addresses composed image retrieval in the context of domain conversion, where the content of a query image is retrieved in the domain specified by the query text. We show that a strong visio...

0123

Nikos Efthymiadis @nikos-efth.bsky.social · 05/12/2024

5/ 📖 arxiv.org/abs/2412.03297 🔗 github.com/NikosEfth/freedom

020

Nikos Efthymiadis @nikos-efth.bsky.social · 05/12/2024

4/ Shout-out to my co-authors for an extremely fruitful collaboration: @billpsomas.bsky.social , Zakaria Laskar, Konstantinos Karantzalos, Yannis Avrithis, Ondrej Chum, @gtolias.bsky.social

120

Nikos Efthymiadis @nikos-efth.bsky.social · 05/12/2024

1) By expanding ImageNet-R to more source domains and by introducing three (3) new benchmarks! 🤯 2) By presenting a novel training-free approach leveraging vision-language models for memory-based textual inversion. 🚀

140

Nikos Efthymiadis @nikos-efth.bsky.social · 05/12/2024

2/ Did you know domain conversion is an underexplored area? Until now, only one dataset (ImageNet-R) existed—and it focused solely on photo-to-rest conversions. Our work expands this field!

020

Nikos Efthymiadis @nikos-efth.bsky.social · 05/12/2024

1/ 🎉 Excited to share our work, "Composed Image Retrieval for Training-Free Domain Conversion", accepted at WACV 2025! 🚀

3165

Nikos Efthymiadis @nikos-efth.bsky.social · 26/11/2024

Hi, vlms are in my research interests, could you add me please?

110

Reposted by Nikos Efthymiadis

Dmytro Mishkin @ducha-aiki.bsky.social · 26/11/2024

Here is a small and cozy starter pack "Prague CV/ML". Most of the people here are not posting much, and I secretly hope that being in the list will motivate them to do so :P go.bsky.app/G9TYiYQ

0249

Nikos Efthymiadis @nikos-efth.bsky.social · 26/11/2024

Hello there! Could you add me please?

110

Reposted by Nikos Efthymiadis

Chris Offner @chrisoffner3d.bsky.social · 21/11/2024

Whenever AI "generates" something impressive, the first question we should always ask is: "What does the closest sample in the training data look like?" LLMs are amazing interfaces for accessing the world's information but they need to be treated as the "search and synthesis" tools they are.

5719

Reposted by Nikos Efthymiadis

Kon Kun 🏳️‍🌈 OPEN FOR WORK @konkun.bsky.social · 21/11/2024

I strongly believe that generative AI is theft and has no place in art. Sadly the beast of generative AI has hurt a lot of professional artists even leaving them without a job. This is something that should concern anyone who values the soul in the art they consume. #artificialinteligence #horror

0102

Reposted by Nikos Efthymiadis

Giorgos Tolias @gtolias.bsky.social · 20/11/2024

Deep global descriptors give a convenient way for retrieval, but local descriptors are a game changer in finding needles in a haystack (particular objects in clutter). Due to their high cost, with AMES we optimize the performance/memory trade-off during re-ranking. #ECCV2024

1328