AI-Based Visual Analysis for Artistic Research

We aim to extend artistic research methodologies by integrating computer vision techniques with knowledge graph-based semantic inference within the WP3 of the EAR project.

Computational Analysis of Artistic Creation Processes

Developing a computational framework to deconstruct both the artistic method of figurative painters and the constructed meaning of their paintings.

Reconstruction and Digital Restoration of Fragmented Artworks

Developing algorithms to reassemble fragmented fresco pieces automatically, such as those from Pompeii. We tackle the challenge of "heavy-tailed" distributions where fragments are severely damaged, eroded, or missing.

My research specifically focuses on:

• Extraction of effective descriptors for irregular and highly damaged fragments,
• Automatic puzzle-solving algorithms combined with human-in-the-loop strategies,
• Inpainting methodologies for the complete digital restoration of missing content.

Computational Numismatics & Visual Analysis

We are establishing a computational framework to address the unique challenges of ancient numismatics. Our work targets visual data heavily affected by centuries of degradation and the inherent variability of manual manufacturing.

Our research specifically focuses on:

• Designing robust recognition algorithms to handle high degradation factors such as surface wear, corrosion (patina), and fractures,
• Computational modeling of stylistic variations resulting from the manual production process (hand-carved dies),
• Die analysis and fine-grained classification to distinguish subtle differences between coin types and individual dies.