Forschung · Publikationen

Unsere Forschung, veröffentlicht.

Die vollständige Liste mit 29 Arbeiten zu Dokumentenanalyse, Handschrifterkennung und neuronalen Netzen.

1. 2026 DCA P0 · Foundations: the biological grounding (four pillars) Preprint (Zenodo) · Neural Computation
2. 2026 DCA P1 · Theory I: the convergence framework (Axioms A1 to A5, Prop. 1 to 4) Preprint (Zenodo) · TMLR
3. 2026 DocVQA 2026 Technical Report · first place (>35B, 60.00%) Preprint (Zenodo) · IJDAR Welf Wustlich nimmt auf der ICDAR 2026 in Wien die Auszeichnung für den ersten Platz bei DocVQA 2026 entgegen.
4. 2026 Enforcing 3D Coherence in Semi-Supervised Segmentation for Pancreatic Tumor Histopathology from Light Sheet Fluorescence Microscopy MIDL 2026 · PMLR 315 · Poster Yousif Hashisho stellt das Paper auf der MIDL 2026 in Taipeh vor. Es ist im Projekt CAPTAIN mit der Universitätsmedizin Göttingen entstanden.
5. 2025 Dynamics of an estuarine biotic community captured in high spatio-temporal resolution using metabarcoding Scientific Data (Nature)
6. 2025 A benchmark analysis of feature selection and machine learning methods for environmental metabarcoding datasets CSBJ
7. 2022 Application of a Deep Learning Approach to Analyze Large-Scale MRI Data of the Spine Healthcare / MDPI
8. 2022 Rescoring Sequence-to-Sequence Models for Text Line Recognition with CTC-Prefixes DAS 2022
9. 2021 Optimizing Small BERTs Trained for German NER Information / MDPI
10. 2021 Mixed Model OCR Training on Historical Latin Script for Out-of-the-Box Recognition and Fine-tuning arXiv
11. 2021 Transformer for Handwritten Text Recognition using Bidirectional Post-Decoding ICDAR 2021
12. 2021 One-Model Ensemble-Learning for Text Recognition of Historical Printings ICDAR 2021
13. 2021 tfaip: a Generic and Powerful Research Framework for Deep Learning based on Tensorflow JOSS
14. 2020 Two Semi-Supervised Training Approaches for Automated Text Recognition
15. 2019 Evaluating Sequence-to-Sequence Models for Handwritten Text Recognition ICDAR 2019
16. 2019 End-To-End Measure for Text Recognition ICDAR 2019
17. 2018 System Description of CITlab's Recognition & Retrieval Engine for ICDAR2017 Competition on Information Extraction in Historical Handwritten Records ICDAR 2017
18. 2018 Neural Text Line Extraction in Historical Documents: A Two-Stage Clustering Approach
19. 2018 READ-BAD: A New Dataset and Evaluation Scheme for Baseline Detection in Archival Documents DAS 2018
20. 2017 A Robust and Binarization-Free Approach for Text Line Detection in Historical Documents ICDAR 2017
21. 2017 Decoding the Output of Neural Networks: A Discriminative Approach
22. 2016 CITlab ARGUS for Keyword Search in Historical Handwritten Documents · ImageCLEF 2016 CEUR
23. 2016 Cells in Multidimensional Recurrent Neural Networks JMLR
24. 2016 Regular Expressions for Decoding of Neural Network Outputs Neural Networks
25. 2016 CITlab ARGUS for Historical Handwritten Documents: HTRtS 2015 (tranScriptorium Dataset)
26. 2014 CITlab ARGUS for Historical Handwritten Documents: HTRtS 2014 Handwritten Text Recognition Task
27. 2014 CITlab ARGUS for Historical Data Tables: ANWRESH-2014 Word Recognition Task
28. 2014 CITlab ARGUS for Arabic Handwriting: OpenHaRT 2013 Document Image Recognition Task
29. 2012 Design Strategies for Weight Matrices of Echo State Networks Neural Computation