Campus Research Institute, Bucharest (2026)

07/09/2026

We – the AIMultimediaLab, coordinated by Prof. Bogdan Ionescu – are excited to share our latest research in Artificial Intelligence and Deep Learning for the diagnosis and staging of membranous glomerulonephritis from electron microscopy images.

In this work, we developed a two-stage Deep Learning model based on vision transformer networks to detect membrane regions and subsequently classify them according to the stage of membranous glomerulonephritis, a major cause of nephrotic syndrome in adults. The proposed model achieved an accuracy of 92.31%, sensitivity of 92.40%, and specificity of 91.67%, while demonstrating strong performance on an independent external dataset. The results highlight the model’s robustness and generalizability, demonstrating the potential of AI-assisted electron microscopy to support more accurate and interpretable MN staging and its integration into clinical diagnostic workflows.

See “Two-Stage Deep Learning Networks for Diagnosing and Staging Membranous Glomerulonephritis from Electron Microscopy Images”, M.G. Constantin, G. Terinte-Balcan, I.M. Lambrescu, T.E. Fertig, O.B. Lazăr, A.-G. Andrei, N. Petre, B. Ionescu, and M. Gherghiceanu, Laboratory Investigation, 106(8), Impact Factor 4.1, 2026. Read the full article here: https://doi.org/10.1016/j.labinv.2026.106141.
Universitatea POLITEHNICA din București Universitatea de Medicina si Farmacie "Carol Davila" Institutul Național Victor Babeș

04/09/2026

We – the AIMultimediaLab, coordinated by Prof. Bogdan Ionescu – are excited to share our latest research on AI-based semantic video indexing, exploring how Artificial Intelligence can transform audiovisual content into searchable and interpretable semantic representations.

In this work, we provide a unified survey of the semantic video indexing pipeline, from temporal segmentation and multimodal information fusion to semantic indexing, video-text retrieval, and knowledge-oriented access. The study examines recent advances in video representation learning, video-language pretraining, audio and speech modeling, dense captioning, and Large Multimodal Models.

The results highlight the importance of approaching semantic video indexing as an end-to-end process that integrates temporal segmentation, representation learning, multimodal fusion, semantic index construction, retrieval, and reasoning. Such reliable and interpretable indexing architectures can enable more effective access to large-scale video collections across applications including media archives, education, healthcare, cultural heritage, and scientific video analysis.

See “AI-Based Semantic Video Indexing: A Unified Survey from Segmentation to Retrieval”, M. Modrogan, R. Țapu, and B. Mocanu, Knowledge-Based Systems, 351, Part B, Impact Factor 8, 2026. Read the full article here: https://doi.org/10.1016/j.knosys.2026.116714.
Universitatea POLITEHNICA din București Institut Polytechnique de Paris

03/09/2026

We – the AIMultimediaLab, coordinated by Prof. Bogdan Ionescu – are excited to share our latest research on computational memorability and the prediction of visual content that leaves a lasting impression on viewers.

In this work, we review the evolution of the MediaEval Predicting Video Memorability benchmark across its 2018–2023 editions, providing a comprehensive overview of the methods, datasets, and evaluation approaches used to predict video memorability. The results highlight the benchmark’s role in advancing reproducible research in computer vision while identifying key challenges that remain, including model interpretability and the subjective, context-dependent nature of human memorability.

See “A Review of Computational Memorability: A Benchmark Framework”, M.G. Constantin, C.-H. Demarty, C. Fosco, S. Halder, G. Healy, B. Ionescu, Ș.V. Luncanu, I. Martín-Fernández, A. Matran-Fernández, R.S. Kiziltepe, A.F. Smeaton, L.-D. Ștefan, L. Sweeney, and A.G. Seco de Herrera, International Journal of Computer Vision, 134, IF 10.3, 2026. Read the full article here: https://doi.org/10.1007/s11263-026-02880-6
Universitatea POLITEHNICA din București University of Cambridge University of Essex Universidad Politécnica de Madrid (Oficial) Ankara Üniversitesi

02/09/2026

We – the Photon-X Spectrum Lab, coordinated by Dr. Ștefan G. Stanciu – are excited to share our latest research on image quality assessment for advanced SHG microscopy, addressing the measure of an image quality when there is no ground truth.

In this work, we evaluate state-of-the-art no-reference image quality assessment (NR-IQA) metrics for second-harmonic generation (SHG) and super-resolved re-scan SHG microscopy. The results show that commonly used AI- and statistics-based image quality metrics often capture competing aspects of microscopy image quality, highlighting the need for complementary evaluation strategies when optimizing image reconstruction, deconvolution, and super-resolution workflows.

See “No-Reference Image Quality Assessment for SHG and Re-scan SHG Microscopy: A Comparative Study”, S.R. Anton, R. Hristu, D.E. Trancă, G.A. Stanciu, M. Popescu, A. Dumitru, and S.G. Stanciu, IEEE Photonics Journal, 18(3), Impact Factor 2.6, 2026. Read the full article here: https://ieeexplore.ieee.org/abstract/document/11505769.
Universitatea POLITEHNICA din București

01/09/2026

We – the AIMultimediaLab, coordinated by Prof. Bogdan Ionescu – are excited to share our latest research on Large Language Model LLMs abstention in healthcare, addressing a critical question for the safe use of AI in clinical settings: when is it safer for an AI model not to answer?

In this work, we reviewed existing approaches to LLM abstention in healthcare, focusing on two key motivations: uncertainty-driven abstention, when a model withholds an answer due to insufficient confidence, and safety-driven abstention, when responding could potentially cause harm. The study shows that state-of-the-art LLMs still struggles with important challenges in refusing unsafe or uncertain medical prompts.

See “When Silence is Safer: A Review and Decision-theoretic Framework for LLM Abstention in Healthcare”, O. Presacan, A. Nik, J. Ojha, V. Thambawita, B. Ionescu, and M.A. Riegler, Nature npj Digital Medicine, Impact Factor 12.4, 2026. Read the full article here: https://doi.org/10.1038/s41746-026-02882-1
Universitatea POLITEHNICA din București Simula Research Laboratory

31/08/2026

We – the Smart and Autonomous Systems Lab, coordinated by Prof. Grigore Cican – are excited to share our latest research on Machine Learning approaches for improving photovoltaic energy prediction and supporting the development of green and sustainable airports in Romania.

In this work, we investigate how advanced Machine Learning/Deep Learning models, including DNN, SRNN, LSTM, GRU, and CNN architectures, can improve photovoltaic energy forecasting using national-level datasets. After evaluating 20,401 unique model configurations, the CNN model achieved the highest predictive accuracy, demonstrating the potential of intelligent forecasting systems for optimizing renewable energy integration in airport infrastructures.

See “Machine Learning for Improving PV Energy Prediction to Develop Green and Sustainable Airports in Romania”, G. Cican, V. Silivestru, F.V. Popescu, and B. Adrian, Engineering Research Express, 8(6), Impact Factor 1.9, 2026. Read the full article here: https://iopscience.iop.org/article/10.1088/2631-8695/ae4db1?utm_source=researchgate.net&utm_medium=article.
Universitatea POLITEHNICA din București Comoti - Romanian Research & Development Institute for Gas Turbines Universitatea Nationala de Aparare "Carol I" - www.unap.ro

07/08/2026

We were delighted to welcome the participants from the 17th European Summer School on Information Retrieval ESSIR2026 for visiting some of our research labs and identifying new opportunities for Information Retrieval technologies to be applied to various engineering fields.

ESSIR is held regularly, providing high-quality teaching of Information Retrieval (IR) and advanced IR topics to an audience of researchers and research students. The mission of the school is to enable students and young professionals to learn about modern research challenges and methods on IR and related disciplines; to stimulate scientific research and collaboration in these fields; and to grow a community of researchers, students, and industry professionals working on IR with collaborations all around the world.

We had the opportunity to showcase some of the research activities carried out within the CAMPUS Research Institute, namely multidisciplinary research in advanced polymers and chemistry, robotics, environmental intelligence, audio technologies, Artificial Intelligence, and aerospace engineering, featuring the Advanced Polymer Materials Group, coordinated by Prof. Iuliana Biru, the Motion Planning for Small Scale Robotics Lab, coordinated by Prof. Florin Stoican, the GEOspatial and Smart SENSors for Environmental Applications Research Lab, coordinated by Prof. Daniela Faur and Prof. Dragos Ioan Sacaleanu, the Spatial Audio Innovation Lab, coordinated by Prof. Victor Popa, the AIMultimediaLab, coordinated by Prof. Bogdan Ionescu, and the Students2SPACE, coordinated by Cătălin Chelmuș.

Through initiatives like ESSIR 2026, we remain committed to strengthening international collaboration, fostering knowledge transfer, and empowering the next generation of researchers and innovators.
Universitatea POLITEHNICA din București

06/08/2026

We – the NOVA Student Lab for UAV/UAS, coordinated by Octavian Gheorghe Balog & Ioan Marian Corjuc – were there, namely at the Air Cargo Challenge 2026, hosted by the University of Stuttgart, in Germany.

Bringing together university teams from across Europe, the competition provided an outstanding platform for students and researchers to demonstrate their engineering skills, innovation, and teamwork in the field of aerospace engineering. The event combined technical presentations, aircraft inspections, and flight demonstrations, challenging participants to design, build, and operate radio-controlled aircraft capable of carrying the maximum possible payload while meeting strict engineering requirements.

We had the chance to participate with our newest flight platform prototype, sharing know-how with other participants and getting feedback from renewed senior engineers. It was a also a great opportunity for building future cooperation opportunities.

Through participation in international competitions such as , we remain committed to supporting excellence in engineering education, fostering innovation, and creating valuable opportunities for students to engage with the international research and academic community.
Universitatea POLITEHNICA din București

05/08/2026

What a treat 😊! During 5 full days, we organized the 17th European Summer School on Information Retrieval ESSIR2026. This year’s event was a coordinated effort between the AIMultimediaLab, CAMPUS Research Institute, Universitatea POLITEHNICA din București, represented by Prof. Bogdan Ionescu, together with Vienna University of Technology, Austria, represented by Dr. Florina Piroi, and CEA LIST, France, represented by Dr. Adrian Popescu, acting as General and Program Chairs.

The participants around the world had the opportunity to explore a wealth of hands-on and lectures on various topics related to Information Retrieval and Artificial Intelligence, namely Introductions to IR, IR Evaluation, Domain-specific IR, Symbolic AI and Knowledge Graphs, Temporal Question Answering, Graph Neural Networks, Explainable AI and IR, Data management for IR, Injecting Context into LLMs, Large Model Biases in IR, Safety of LLMs in IR, RAG and Generative Retrieval, Brain Machine Interfaces and IR, and Agentic RAG. In addition, they completed a hackathon on LLMs design, participated in the 14th PhD Symposium on Future Directions in Information Access, and visited the CAMPUS Research Center labs. The Romanian cuisine was represented at its finest via a Welcome Reception at the Zaganu craft beer bar in Bucharest’s new town, and a Gala Dinner at Hanu’ lui Manuc in Bucharest’s old town.

Many thanks to our speakers: Prof. Henning Müller, HES-SO & Université de Genève, Switzerland, Prof. Iadh Ounis, University of Glasgow, United Kingdom, Prof. Marta Sabou, Vienna University of Economics and Business, Austria, Prof. Adam Jatowt, Universität Innsbruck, Austria, Prof. Gabriella Pasi, Università degli Studi di Milano-Bicocca, Milano, Italy, Prof. Yashar Moshfeghi, University of Strathclyde, United Kingdom, Prof. Wojciech Kusa, NASK National Research Institute, Poland, Prof. Udo Kruschwitz, Universität Regensburg, Germany, Dr. Svitlana Vakulenko, WU (Wirtschaftsuniversität Wien), Austria, Dr. Florina Piroi, TU Wien, CRDM, Austria, Dr. Adrian Popescu, Université Paris-Saclay, CEA LIST, France, Lucian Gruia, Ciklum & AI Multimedia Lab, Romania, Octavian Radu, AI Multimedia Lab & London Stock Exchange, Romania, Dr. Anirban Chakraborty, University of Wolverhampton, United Kingdom.

Many thanks to our participants without whom, ESSIR 2026 would not have been possible!

Many thanks to all the organizing teams, including the local organizing team and speakers!

We would like to also acknowledge our sponsors, namely ACM via the SIGIR group (travel grants), Siemens Industry Software Romania, SAP Romania, and VOIS Romania.

We hope you enjoyed the experience as much as we did! We are looking forward already to next year’s edition! Through initiatives like this, we are proud to support internationalization, training students and young professionals, know-how transfer and collaboration around the world.

04/08/2026

We – the National Center for Micro and Nanomaterials, represented by Dr. Bogdan Vasile – were happy to organize the 2nd working workshop of the project PREWA “Sustainable Phosphorus Recovery from Waste through Integrated Environmental and Energy Management in the Cleaner Production Chain”, and welcome our partners from the Solar Energy Institute, Ege Üniversitesi, Türkiye, led by Prof. Özben Kutlu.

The meeting provided a valuable opportunity to review the progress achieved so far, exchange expertise, coordinate upcoming activities, and further strengthen the collaboration between the Romanian and Turkish research teams.

Through international partnerships such as Project PREWA, we remain committed to advancing interdisciplinary research and developing sustainable solutions for resource recovery, environmental protection, and cleaner production.
Universitatea POLITEHNICA din București

Campus Research Institute

07/09/2026

04/09/2026

03/09/2026

02/09/2026

01/09/2026

31/08/2026

07/08/2026

06/08/2026

05/08/2026

04/08/2026

Address

Website

Alerts

Contact The University

Shortcuts

Share

Category

Listing takedown request