Welcome from MLA²S

• Patrick McAllister (Institute for the Cultural and Intellectual History of Asia (IKGA) / Cluster of Excellence ‘EurAsian Transformations’)
• Claus Trost (Erich Schmid Institute of Materials Science of theAustrian Academy of Sciences)
• Jan Odstrčilík (Institut für Mittelalterforschung, ÖAW)
• Kati Heinrich (IGF | ÖAW)
• Nicki Holighaus (Acoustics Research Institute, Austrian Academy of Sciences)
• Claudius Krause (HEPHY Vienna (ÖAW))

Room: Seminarraum 7/8 Welcome with coffee

Introductory Remarks

• Patrick McAllister (Institute for the Cultural and Intellectual History of Asia (IKGA) / Cluster of Excellence ‘EurAsian Transformations’)

Room: Seminarraum 7/8

Example Images from IHB and ÖAI

• Stephan Kurz (IHB)
• Helmut Schwaiger (Austrian Academy of Sciences / Austrian Archaeological Institute)

Room: Seminarraum 7/8

Image Acquisition and Enhancement in Cultural Heritage

• Florian Kleber (Computer Vision Lab, TU Vienna)
• Christian Stippel (Computer Vision Lab, TU Vienna)

Room: Seminarraum 7/8 Talk with Discussion. Historical photographs and archival images often degrade due to noise, blur, fading, low resolution, or physical damage, which poses significant challenges for acquisition, analysis, preservation, and accessibility. The first part of the lecture deals with image acquisition using multispectral imaging systems. This is mainly used for manuscripts to improve readability or to make palimpsest texts visible again. The second part of this talk examines modern deep learning techniques for image restoration, with a focus on CNN-, transformer-, and autoencoder-based models. We compare their performance across key tasks such as denoising, deblurring, super-resolution, and inpainting, highlighting strengths, limitations, and trade-offs. Real-world case studies illustrate how these methods can recover lost details while maintaining authenticity. Finally, we present a web-based demonstration that allows participants to upload and restore their own images interactively.

Bayesian Imaging with Learned Priors

• Martin Holler (IDea_Lab, Uni Graz)

Room: Seminarraum 7/8 Talk with Discussion. Motivated by concrete applications in image reconstruction, I will present an introduction to Bayesian imaging and related machine learning approaches. In particular, I will present some basics about probabilistic machine learning and the underlying statistical notions and concepts. Then I will discuss the learning of images priors and applications in biomedical imaging. Literature: https://arxiv.org/abs/2507.12432

Biological image analysis in the age of AI - an overview and our experience in histopathology

• André Rendeiro (CeMM - Research Center for Molecular Medicine)

Room: Seminarraum 7/8 Talk with Discussion. My talk explores the expanding landscape of biomedical image analysis, from the macro-scale of radiology to the micro-scale of histopathology and microscopy. I will discuss the unique challenges presented by these diverse imaging modalities, including high dimensionality, noise, and complex spatial relationships. I will highlight the application of deep learning-based methods for image restoration in two key areas: denoising and super-resolution. I will illustrate how these techniques can recover biological information from low-quality/low-resolution data. In my group, we also develop methods to address specific challenges in histopathology, leveraging large-scale datasets and deep learning to decipher the organization of tissues in the human body. Finally, I will discuss the crucial role of the open-source community in democratizing access to advanced image tools, with an emphasis on code quality, reproducibility, and documentation, and support. These are essential for making tools developed by some adopted, used and eventually transform a whole field.