Special Sessions

Image Quality Assessment and Enhancement in the Context of Medical Imaging and Diagnosis

  • Hantao Liu (Cardiff University, United Kingdom)
  • Lu Zhang (National Institute of Applied Sciences (INSA) of Rennes, France)
  • Azeddine Beghdadi (University Sorbonne Paris Nord, France)

This special session aims at collecting contributions of new approaches to understand and estimate image quality in the context of medical imaging, diagnosis and surgery, as well as image processing methods (denoising, reconstruction, etc.) to enable image quality enhancement for a better diagnostic or surgery task performance. More specifically, we are interested in models and techniques that tackle the challenges of subjective protocol and objective quality estimation of medical images and videos.

Topics of interest include, but are not limited to:

  • Publicly accessible datasets for medical image quality assessment
  • New test protocol proposals for evaluating medical image quality, related to diagnostic or surgery tasks
  • Experiments for understanding the image quality interpretation in medical context
  • Objective models for estimating the medical image quality
  • Learning-based methods for predicting the medical image quality
  • Quality assurance of Image-guided radiotherapy
  • Approaches modeling a diagnostic process from medical experts’ point of view
  • Image/video quality assessment for intelligent emergency systems
  • Human visual attention experiments and prediction models for a better image quality understanding or modeling
  • Quality enhancement methods in the medical context
  • Quality related medical image processing and applications

AI in the City: Efficient, Scalable and Privacy-Preserving Visual Scene Understanding in Man-Made Environments

  • Amine Bourki (VizioSense AI Lab, Paris, France)
  • Mohib Ullah (NTNU, Gjøvik, Norway)

Man-made environments such as buildings and cities are typically characterized by their large scale, dynamic nature and the dense population of objects and people they contain. With the recent significant advances in Visual AI and the democratization of massive visual information that enable autonomous systems, surveillance applications, and real-time data acquisition from indoor or outdoor man-made scenes, the processing of such visual information requires a specific attention to several aspects. These include robustness, data- and computation efficiency, scalability, as well as addressing the ethical concerns associated with individual privacy which can be compromised through direct (e.g., facial recognition), or indirect (e.g., vehicle plate recognition) identification methods. This special session aims to encourage novel contributions in the field of visual scene understanding, with an emphasis on addressing the challenges related to robustness, scalability, privacy and dynamic scenes in the realm of large, dynamic, densely populated scenes such as cities and buildings.


Topics of interest include, but are not limited to:

Scene understanding:

  • Semantic, instance level and panoptic segmentation,
  • Object detection,
  • Joint geometric-semantic reasoning,
  • 3D from multi-view and sensors,
  • Efficient and scalable vision,
  • Segmentation and shape analysis,
  • Perceptual grouping and counting,
  • Crowd analysis,
  • Behavior analysis,
  • Multi-modality, RGB-D and point clouds,
  • Drones and autonomous vehicles.

  • Dynamic environments:
  • Multi-Object Tracking (MOT),
  • Person and object re-IDentification (re-ID),
  • Multi-Camera, Multi-Object Tracking (MC-MOT),
  • Real-time acquisition and processing,
  • Object pose estimation.

  • Privacy, Edge AI and social impact:
  • Privacy-by-design in visual processing systems,
  • Embedded and video surveillance systems,
  • Deep Learning for compute-constrained and low-power devices,
  • Power efficient Deep Learning and Green AI,
  • Ethics, fairness and explainability in visual AI.

  • Research reproducibility:
  • Datasets and evaluation.

  • Spectral Imaging and its Applications

    • Sony George (NTNU Norway)

    Spectral imaging, an advanced imaging technique that brings together the advantages of conventional digital imaging, as well as the power of spectroscopy. Originally developed in remote sensing domain, but later find its application in several areas such as food, forensics, cultural heritage, medical, etc, to list a few. There is lot of growing interest of Spectral imaging both at the reasearch and industry levels.

    The proposed special session in EUVIP conference intends to cover the research in the following areas, but not limited to, the topic of spectral imaging:

  • Multi/hyperspectral imaging
  • Spectral imaging sensors- design and calibration
  • Physical modeling
  • Spectral image processing
  • Classification, unmixing, segmentation
  • Application of spectral imaging- Remote sensing, cultural heritage, food, medical etc.
  • Deep learning and artificial intelligence for spectral image processing
  • Please use the Abstract Template provided for the EUVIP conference.
    The deadline for abstract submission is June 05, 2023.