Nikos Ypsilantis

About me

I am a Ph.D. candidate and a researcher at the Visual Recognition Group, Czech Technical University in Prague. My supervisor is Professor Ondrej Chum.

My research lies in the intersection of computer vision and deep learning, and more specifically I focus on image representation learning for fine-grained image recognition tasks. I am mostly interested in developing large-scale image retrieval systems that work on the finest of granularities, down to the instance-level, and across multiple visual domains.

Prior to my Ph.D., I obtained my Engineering Diploma (5 year degree) at the school of Electrical & Computer Engineering, National Technical University of Athens.

Academic/Research News

[Feb. 25] : Our ILIAS dataset is now accepted at CVPR 2025! It is a new large-scale test dataset for instance-level image retrieval, designed to support future research in image-to-image and text-to-image retrieval for particular objects. It is the outcome of a large effort from researchers from the Visual Recognition Group, with whom I am very grateful to have collaborated!

[Dec. 24] : I attended the NeurIPS 2024 conference in Vancouver, Canada. There, I presented our work on Universal Dynamic Online Distillation [4]. It was a great experience with a lot of interesting talks and discussions!

[Sep. 24] : Our paper titled "Universal Dynamic Online Distillation for generic image representations" [4] was accepted at NeurIPS'24. Special thanks to my advisor and our collaborators Kaifeng Chen and Andre Araujo from Google!

[Sep. 24] : Our paper : Co-Segmentation without any Pixel-level Supervision with Application to Large-Scale Sketch Classification [3] was accepted at ACCV'24! Special thanks to my advisor!

[July. 24] : I attended the International Computer Vision Summer School (ICVSS) 2024, which took place in Sicily, Italy!

[Feb. 24] : I attended and gave an invited talk to Computer Vision Winter Workshop 2024, which took place in Terme Olimia, Slovenia!

[Oct. 23] : I attended the International Conference on Computer Vision in Paris, where I presented our work on Universal Image Embeddings [2]. It was a very rewarding experience after a lot of months of hard work!

[Jul. 23] : Our paper on Universal Image Embeddings [2] was accepted at ICCV'23. Special thanks to my advisor and the great team of researchers from Google, especially Kaifeng Chen and Andre Araujo for the great collaboration!

[Jul. 22] : I attended the Vision and Sports Summer School in Prague! Fantastic experience with interesting lectures, fun sports and a lot of new friends!

[Jan. 22] : Exciting news! I was accepted to start my Ph.D. studies at the Visual Recognition Group, Department of Cybernetics, Faculty of Electrical Engineering, Czech Technical University in Prague. I will be working on large-scale visual recognition under the supervision of Professor Ondrej Chum.

[Oct. 21] : I presented the Met Dataset [1] at the 4th Instance-Level Recognition Workshop of ICCV'21. Here is the video of the presentation.

[Oct. 21] : My first paper [1] got accepted at NeurIPS 2021 Track on Datasets and Benchmarks! Special thanks to my supervisor Giorgos Tolias and the rest of the team members for this amazing collaboration!

Publications

During Ph.D.

[4]: Nikolaos-Antonios Ypsilantis, Kaifeng Chen, André Araujo, Ondřej Chum: "UDON: Universal Dynamic Online distillatioN for generic image representations", Neural Information Processing Systems (NeurIPS) 2024 [arXiv]

[3]: Nikolaos-Antonios Ypsilantis, Ondřej Chum: "Co-Segmentation without any Pixel-level Supervision with Application to Large-Scale Sketch Classification", Asian Conference on Computer Vision (ACCV) 2024 [arXiv]

[2]: Nikolaos-Antonios Ypsilantis, Kaifeng Chen, Bingyi Cao, Mário Lipovský, Pelin Dogan-Schönberger, Grzegorz Makosa, Boris Bluntschli, Mojtaba Seyedhosseini, Ondřej Chum, André Araujo: "Towards Universal Image Embeddings: A Large-Scale Dataset and Challenge for Generic Image Representations", International Conference on Computer Vision (ICCV) 2023 [project page]

Before Ph.D.

[1]: Nikolaos-Antonios Ypsilantis, Noa Garcia, Guangxing Han, Sarah Ibrahimi, Nanne Van Noord, Giorgos Tolias: "The Met Dataset: Instance-level Recognition for Artworks", Neural Information Processing Systems (NeurIPS) 2021, Datasets and Benchmarks Track [project page]

Website hosted on GitHub Pages
Theme adapted from orderedlist/minimal.