AI & Computer Vision Research Hub

Led by Dr. Silvia Cascianelli

Welcome to our research hub, where we explore advanced techniques in image generation, handwriting imitation, and document understanding in a collaborative, mentoring environment.

Projects

Image Generation

• Shifting the Breaking Point of Flow Matching for Multi-Instance Editing (IDAttn) [arxiv:2602.08749v1]
• Merging and Splitting Diffusion Paths for Panoramas [arXiv:2408.15660]
• Alfie: Democratising RGBA Image Generation [arXiv:2408.14826]

Handwriting Imitation

• Autoregressive Styled Text Image Generation, but Make it Reliable (Eruku) [arxiv:2510.23240]
• Zero-Shot Styled Text Image Generation, but Make It Autoregressive (Emuru) [arxiv:2503.17074]
• VATr++: Handwritten Text Generation [arXiv:2402.10798]
• Handwritten Text Generation from Visual Archetypes [arXiv:2303.15269]
• HWD: Evaluation for Handwritten Text Generation [arXiv:2310.20316]
• Pretrained HTR Models for Fine-Tuning [arXiv:2305.02593]

Document Understanding

• μgat: Document Parsing with Multi-Page Context [arXiv:2408.15646]
• Volumetric FFC for Ink Detection on Papyri [arXiv:2308.05070]
• Synthetic Pre-Training for Handwriting Processing [arXiv:2304.01842]
• The LAM Dataset for Handwritten Text Recognition [arXiv:2208.08109]
• Boosting HTR with Deformable Convolutions [arXiv:2305.02593]

Our Team

• Silvia Cascianelli – Assistant Professor
• Vittorio Pippi – PhD Student
• Fabio Quattrini – PhD Student
• Carmine Zaccagnino – Research Intern
• Pau Torras Coloma – PhD Student
• Bram Vanherle – CV Engineer

Contact

Email: scascianelli[at]unimore[dot]it
LinkedIn: Silvia Cascianelli