Prof. Dr. Vasileios Belagiannis

Lehrstuhl für Multimediakommunikation und Signalverarbeitung

Our research focuses on basic and applied research in machine learning.  We develop approaches for anomaly detection, uncertainty estimation, out-of-distribution detection, few-shot learning, noisy label learning, as well as hardware-aware algorithms for model compression and efficient design of neural network architectures. Our applications include medical image analysis with problems such as cell counting and segmentation in histology images, organ segmentation from CT and MRI volumes, and multi-label medical image segmentation.

Research projects

  • Few-shot learning for medical images.
  • Cell Segmentation.
  • Multi-label image classification.

  • Subcontracting within the framework of the ÖGP NXT GEN AI METHODS – Generative methods for perception, prediction, and planning (NXTAIM)

    (Third Party Funds Single)

    Project leader:
    Term: 1. January 2026 - 31. December 2026
    Acronym: ÖGP NXT GEN AI METHODS
    Funding source: Industrie
    URL: https://nxtaim.de/en/home/

    The aim of this project is to develop self-playing, multi-agent simulators based on GPUDrive. In this context, we will develop trajectory planning strategies with a focus on efficient training and optimisation. These strategies will then be tested in automated driving scenarios. 

  • Bavarian Advanced Resolution Radar

    (Third Party Funds Group – Sub project)

    Overall project: Bavarian Advanced Resolution Radar
    Project leader:
    Term: 1. February 2025 - 31. January 2028
    Acronym: BAVAR-RADAR
    Funding source: Bayerisches Staatsministerium für Wirtschaft, Landesentwicklung und Energie (StMWi) (seit 2018)
  • SUSTAINET-inNOvAte: Nachhaltige Technologien für fortschrittliche resiliente und energieeffiziente Netze - Reibungslose, sichere und widerstandsfähige Netze für die dynamische digitale Welt

    (Third Party Funds Group – Sub project)

    Overall project: Sustainable Technologies for Advanced Resilient and Energy-Efficient Networks - Frictionless, secure, and resilient communication networks for the dynamic digital world
    Project leader:
    Term: 1. January 2025 - 31. December 2027
    Acronym: SUSTAINET-inNOvAte
    Funding source: BMFTR / Verbundprojekt
  • Subcontracting within the framework of the ÖGP NXT-AIM Generative Modeling

    (Third Party Funds Single)

    Project leader:
    Term: 1. January 2024 - 31. December 2026
    Acronym: Unterbeauftragung ÖGP NXT-AIM
    Funding source: Industrie
    URL: https://nxtaim.de/en/home/

    The project deals with two aspects of generative modeling. First, generative models, especially fundamental models, have made a significant contribution in the areas of image, text, and audio. However, they have not yet been well researched for sequential and unstructured data, such as automotive data. Second, the latent spatial representation in generative models is not interpretable. However, this is linked to the predicted or generated output. This project aims to explore generative models for trajectory planning by integrating robustness measures. 

2026

2025

2024

2023

2022

2021

2020

Related Research Fields:

Contact: