Wolfgang Fuhl
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Wolfgang Fuhl has a total of 63 co-authors. Here are the top co-authors by number of shared publications:
- Enkelejda Kasneci
Technical University of Munich, Munich, Germany and Munich Center for Machine Learning (MCML), Munich, Germany
22 shared publications - Kay Nieselt
Institute for Bioinformatics and Medical Informatics, University of Tübingen, Tübingen, Germany
7 shared publications - Thiago Santini
University of Tübingen Sand 14 -72076 Tübingen Germany
7 shared publications - Anne Herrmann-Werner
Tübingen Institute for Medical Education, University of Tübingen, Tübingen, Germany
6 shared publications - Daniel Weber
Department of Human Computer Interaction University Tübingen Tübingen 72076
6 shared publications Enkelejda Kasneci
University Tübingen
6 shared publications
Latest Publications
- Open Access
- 30 September 2026
- Published by MDPI AG in Preprints.org
A Novel U-Net Modification Inspired by Self-Attention for Image Segmentation
While the transformer architecture is currently the most powerful machine learning architecture, it requires massive amounts of data for training. In addition, the execution requires large amounts of memory since the self attention increases the model size quadratically based on the amount of key, query, and value assignments. The U-Net architecture, which is based on normal convolutions, batch normalization, and residual blocks, is a fast and efficient DNN architecture that can effectively be used for semantic segmentation. In this paper, I evaluate different U-Net modifications and compare them to the state of the art as well as the baseline U-Net with cross connections. Additionally, I propose a novel modification that extends the cross connections together with resizing. The proposed approach slows the model slightly but improves the performance significantly. All evaluations are done on the TEyEDS dataset.
- 30 September 2026
- Published by MDPI AG in Preprints.org
Open AccessA Novel U-Net Modification Inspired by Self-Attention for Image Segmentation
While the transformer architecture is currently the most powerful machine learning architecture, it requires massive amounts of data for training. In addition, the execution requires large amounts of memory since the self attention increases the model size quadratically based on the amount of key, query, and value assignments. The U-Net architecture, which is based on normal convolutions, batch normalization, and residual blocks, is a fast and efficient DNN architecture that can effectively be used for semantic segmentation. In this paper, I evaluate different U-Net modifications and compare them to the state of the art as well as the baseline U-Net with cross connections. Additionally, I propose a novel modification that extends the cross connections together with resizing. The proposed approach slows the model slightly but improves the performance significantly. All evaluations are done on the TEyEDS dataset.
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Real Time Detection System of Cataract, Glaucoma and Retinal Disease on a Raspberry Pi 5
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Open AccessReal Time Detection System of Cataract, Glaucoma and Retinal Disease on a Raspberry Pi 5
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
A Deep Learning Approach for the Estimation of Eye Ball Torsion
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Open AccessA Deep Learning Approach for the Estimation of Eye Ball Torsion
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
NERFs for Scanpath Reconstruction and Generation
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Open AccessNERFs for Scanpath Reconstruction and Generation
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Recognition of errors in gaze-based interaction with anomaly detection
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-6
Open AccessRecognition of errors in gaze-based interaction with anomaly detection
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Can We Predict The Next Fixation For Web Based Video Platforms?
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Open AccessCan We Predict The Next Fixation For Web Based Video Platforms?
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Scanpath Classification with an n-mer Deep Neural Network Architecture
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessScanpath Classification with an n-mer Deep Neural Network Architecture
- Open Access
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
An Annotation Tool for the Slow and Fast Phases of Nystagmus, Applicable to Clinical Caloric Testing
- 25 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-2
Open AccessAn Annotation Tool for the Slow and Fast Phases of Nystagmus, Applicable to Clinical Caloric Testing
- 12 May 2025
- Published by Springer Nature
- p. 151-167
Normalized Validity Scores for DNNs in Regression Based Eye Feature Extraction and Real-Time Models for the Raspberry Pi
- 12 May 2025
- Published by Springer Nature
- p. 151-167
Normalized Validity Scores for DNNs in Regression Based Eye Feature Extraction and Real-Time Models for the Raspberry Pi
- 2 December 2024
- Published by Springer Nature
- p. 292-304
A Trainable Feature Extractor Module for Deep Neural Networks and Scanpath Classification
- 2 December 2024
- Published by Springer Nature
- p. 292-304
A Trainable Feature Extractor Module for Deep Neural Networks and Scanpath Classification