Said Gurbuz

Moving Beyond Sparse Grounding with Complete Screen Parsing Supervision

A. Said Gurbuz, Sunghwan Hong, Ahmed Nassar, Marc Pollefeys, Peter W. J. Staar

TL;DR A large-scale dataset and training framework for complete screen understanding — parsing all UI elements with bounding boxes and semantic labels from web screenshots.

Identify, Locate, Link: End-to-End Key-Value Extraction from Document Images

A. Said Gurbuz, A. Nassar, C. Auer, M. Lysak, L. Morin, M. Omenetti, T. Strohmeyer, P. Vagenas, N. Livathinos, M. Dolfi, P. W. J. Staar

TL;DR An end-to-end approach that jointly identifies, locates, and links key-value pairs in visually rich document images without external OCR.

SmolDocling: An Ultra-Compact Vision-Language Model for End-to-End Multi-Modal Document Conversion

A. Nassar, M. Omenetti, M. Lysak, N. Livathinos, C. Auer, L. Morin, R. T. de Lima, Y. Kim, A. Said Gurbuz, M. Dolfi, P. W. J. Staar

TL;DR A 256M-parameter vision-language model that converts complex documents end-to-end using a novel DocTags markup, competing with models 27x larger.

An Empirical Analysis of IDS Approaches in Container Security

Y. Sever, G. Ekinci, A. H. Dogan, B. Alparslan, A. Said Gurbuz, V. Jabrayilov, P. Angin