진학프로석·박사 채용 접수 플랫폼

Czech Technical University in Prague (CTU) offers a fellowship program, the CTU Global Postdoc Fellowship. This new and attractive two-year fellowship-program offers excellent researchers who have recently completed their PhD the chance to continue their research career at CTU. Fellows receive a two year fellowship and become members of a team led by a mentor.

The scholarship applicant must meet the following conditions on the date of application:

• • be no more than 7 years since obtaining the first Ph.D. degree,
• • Ph.D. studies at a university outside the Czech Republic or have completed at least a one-year working research stay abroad (outside the Czech Republic),
• • be an author (co-author) of three or more publications in a journal with IF or CORE A*/A conference paper.

Recent progress in text-to-image generation, fueled by diffusion models and large-scale training, has greatly improved image synthesis aligned with textual prompts. This project, however, focuses on a complementary challenge: generating diverse image variations that preserve the identity and fine-grained appearance of a specific object. The variations may involve changes in background, illumination, viewpoint, or non-rigid deformations. The target object will be defined either by an exemplar image or a detailed textual description. The resulting images will support training and evaluation of instance-level (particular object) recognition systems. In the literature, this line of work falls under personalized generative modeling, where “personalization” refers to conditioning on a specific object instance.

Giorgos Tolias is an associate professor at CTU in Prague and a member of the Visual Recognition Group (VRG). He is currently leading a small team within VRG supported by a Junior Start GACR Grant. From 2014 to mid-2015 he was a postdoc at the LinkMedia team of Inria Rennes, while he obtained my Ph.D from the NTUA, Greece in 2013. He is an associate editor of IJCV and was an area chair for ECCV2020, ECCV2024, and ICCV2025. His work has been published numerous times in top computer vision conferences and journals such as CVPR (x13), ECCV (x6), ICCV (x3), IJCV (x3) and PAMI (x1). His work received a best paper award honorable mention at BMVC’17 (core rank A).

Advising history: postdocs (1+2current), PhD (1+4 current) MSc (13), Bc (5)

Skills/Qualifications

The ideal candidate has a strong research background in generative models, particularly in text-to-image synthesis and/or personalized generation that preserves object identity or style. They should have a solid publication record in top-tier computer vision conferences such as CVPR, ICCV, or ECCV, and demonstrate proficiency in deep learning frameworks (e.g., PyTorch), experience with large-scale datasets, and a collaborative mindset.