LFLeopold Fischer-Brandies

Leopold Fischer-Brandies

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Leopold Fischer-Brandies has a total of 9 co-authors. Here are the top co-authors by number of shared publications:

  • Ricardo Buettner

    Chair of Hybrid Intelligence, Helmut-Schmidt-University/University of the Federal Armed Forces Hamburg, Hamburg, Germany

    7 shared publications
  • Christian Woesle

    Chair of Hybrid Intelligence, Helmut-Schmidt-University/University of the Federal Armed Forces Hamburg, Hamburg, Germany

    2 shared publications
  • Lucas Müller

    Chair of Hybrid Intelligence, Helmut-Schmidt-University/University of the Federal Armed Forces Hamburg, Hamburg, Germany

    2 shared publications
  • Susan Bertram

    Chair of Hybrid Intelligence, Helmut-Schmidt-University/University of the Federal Armed Forces Hamburg, Hamburg, Germany

    2 shared publications
  • Benjamin Rebholz

    Chair of Hybrid Intelligence, Helmut-Schmidt-University/University of the Federal Armed Forces Hamburg, Hamburg, Germany

    1 shared publication
  • Christopher Mai

    Chair of Hybrid Intelligence, Helmut-Schmidt-University/University of the Federal Armed Forces Hamburg, Hamburg, Germany

    1 shared publication

Latest Publications

  • Open Access

    DefectAtlas-20: A real-world image dataset for joint material and surface-condition classification

    by Ricardo Buettner, Leopold Fischer-Brandies, Lukas Pasold, Luca Eisentraut

  • Open Access

    A Robust Occlusion-Aware Deep Learning Architecture for Thermal Aerial Person Classification in Search-and-Rescue Missions

    by Christian Woesle, Leopold Fischer-Brandies, Ricardo Buettner

    Uncrewed aerial vehicles equipped with long-wave infrared cameras are a promising tool for locating missing persons, yet their performance collapses when more than 70% of a human body is obscured by vegetation. We address this limitation by reformulating the task as a thermal classification problem rather than a detection problem, allowing the system to focus on thermal appearance cues that remain visible under heavy occlusion. Recognition robustness is further affected by high-frequency infrared sensor noise, which increases with flight altitude. Using five-fold cross-validation at flight altitudes of 30 m, 50 m, and 70 m, we show that the proposed classification pipeline achieves an accuracy of 99.07% and maintains strong recall under heavy occlusion while operating at a low computational cost. Analysis of altitude-specific models demonstrates that the Gaussian-enhanced pipeline provides a statistically significant improvement in robustness, with the largest gains observed at higher flight altitudes due to more effective suppression of altitude-dependent sensor noise. These findings establish an operational robustness benchmark for occlusion-aware thermal person classification and provide a reproducible foundation for improving the reliability of uncrewed aerial vehicle search-and-rescue systems.


  • Open Access

    The Impact of Combining Datasets on the Robustness of Deep Learning Architectures: A Cross-Dataset Analysis

    by Ricardo Buettner, Susan Bertram, Leopold Fischer-Brandies

    Convolutional Neural Networks have become a widely used technology. Typically, the performance of CNN models is measured using an accuracy score obtained on a single dataset. This often results in systems that perform significantly worse in real-world applications. The robustness of the model to unseen environments is still underresearched. Specifically, a research gap remains on how training data affects the robustness of deep learning systems. This research article investigates the impact of combining data in training datasets on the robustness and performance of deep learning models through a cross-dataset analysis. We employ a transfer learning approach to train deep learning models based on four popular architectures and two different datasets, as well as a combination of both datasets. Our results demonstrate that combining two datasets can improve robustness, but the specific effects on performance can vary between architectures, leading to a slight decrease in accuracy in most observed cases, or even an accuracy gain. Furthermore, we find that training on more complex datasets tends to outperform training on simpler datasets in cross-evaluation settings, indicating that models trained on more complex training datasets are more robust. However, we also observe that a simpler architecture fails to generalize when trained on the combined training data, indicating the need for caution and extensive evaluation when combining datasets during the development cycle of deep learning systems.


  • Top Cited Papers
    Open Access

    A Systematic Literature Review of Hallucinations in Large Language Models

    by Christian Woesle, Leopold Fischer-Brandies, Ricardo Buettner

    This review systematically maps research on hallucinations in large language models using a descriptive scheme that links model outputs to four system architectures: unaugmented generation, post-hoc reactive validation, proactive detection-and-mitigation, and fully integrated detection-and-mitigation designs. Our methodology for this systematic review follows the PRISMA guidelines to ensure transparency and reproducibility. We searched IEEE Xplore, ACM Digital Library, and ScienceDirect for studies published between 2015 and January 2025 and extracted 125 peer-reviewed papers across nine application domains. Quantitative analysis shows that question answering and multimodal tasks account for 48% of all papers, whereas software engineering, educational technology, and autonomous systems are underexplored. Although 87.5% of the studies rely on additional reactive or proactive defenses, only 8.8% implement integrated architecture-level safeguards, revealing a critical gap in unified and dynamic architectures. The resulting classification matrix and domain map provide a diagnostic tool for locating blind spots and comparing architectural maturity. Three actionable priorities emerge: develop integrated reasoning-and-verification loops that pre-empt hallucinations; transfer proven causal-intervention and multi-agent validation pipelines to high-stakes, under-represented domains and benchmark them under real conditions; and build modular, cross-domain evaluation frameworks that isolate the contribution of individual mitigation components and support ablation studies. By consolidating fragmented evidence and quantifying architecture-domain imbalances, this review establishes a traceable foundation for engineering reliable, explainable, and domain-adaptable countermeasures to hallucinations in generative language technology.


  • Open Access

    To Combine or Not to Combine? The Influence of Combining Training Datasets on the Robustness of Deep Learning Models: An Analysis for Optical Character Recognition of Handwriting

    by Leopold Fischer-Brandies, Lucas Müller, Benjamin Rebholz, Ricardo Buettner

    The present manuscript addresses the question of how training data should be sampled for deep learning models by analyzing and evaluating the impact of training data representation and complexity on the performance and robustness of deep learning models. To address this open question, we take a combinatorial approach and train three architecturally identical deep learning models on three combinations of handwritten digit datasets of varying complexity: EMNIST Digits, DIDA, and a newly composed third dataset combining the first two datasets. Each model was evaluated using withheld test data from all three datasets. We find that models trained exclusively on either EMNIST Digits or DIDA performed well on their respective datasets but poorly on unfamiliar datasets. However, the model trained on both datasets showed an overall solid performance, although not quite reaching the accuracy of the specialized models on their respective datasets. We conclude that while specializing in the training dataset can increase accuracy, a more diverse dataset enhances model robustness. In practice, deep learning models should thus be trained on data that represent the actual application environment as closely as possible or, if such data is not available, on diverse data.


  • Open Access

    Fresh or Rotten? Enhancing Rotten Fruit Detection With Deep Learning and Gaussian Filtering

    by Leopold Fischer-Brandies, Lucas Müller, Justus Johannes Riegger, Ricardo Buettner

    We address the pressing issue of food waste by proposing a robust image classification model that can reliably detect rotten fruits. More than half of the fruit yield is lost along the supply chain, with post-harvest losses due to rottenness playing a pivotal role, as even a single decomposing piece can cause huge damage to nearby produce. Our transfer learning-based model uses the ResNet50 convolutional neural network architecture as a binary classification model to distinguish between fresh and rotten fruits. The model performance is enhanced with a Gauss filter and a dropout layer to ensure robustness and prevent overfitting. We achieve high accuracies beyond 99% on unseen test data, setting a new benchmark and outperforming previous efforts. Our work has theoretical and practical implications. To the best of our knowledge we are the first to explore the use of Gauss filter to preprocess input images in fruit classification. We find that Gauss filters with small kernel size improve the performance of our model. Our research can improve post-harvest applications through automation and can thus help reduce food waste, improve food safety, and reduce costs for growers, distributors, and retailers, and thereby improve the overall efficiency of the supply chain.


  • Open Access

    AsphaltCrackNet: A Novel Architecture for Classifying Cracks in Asphalt Pavement

    by Leopold Fischer-Brandies, Susan Bertram, Christopher Mai, Ricardo Buettner