http://dl.gi.de/handle/20.500.12116/37601 2026-07-21T13:21:05Z http://dl.gi.de/handle/20.500.12116/37769 From Physical to Virtual: Leveraging Drone Imagery to Automate Photovoltaic System Maintenance Lowin, Maximilian; Kellner, Domenic; Kohl, Tobias; Mihale-Wilson, Cristina Optimizing the maintenance of large-scale infrastructure can be a significant cost driver for small and medium-sized enterprises (SMEs). This paper presents a feasible approach to combine data from real-world physical structures collected through an automated maintenance process with cloud-based AI services to generate a meaningful virtual representation of such structures. We use photovoltaic systems as an exemplary physical structure and thermal imaging, collected through scheduled drone monitoring. With help of these unstructured data sources, we demonstrate our approach's applicability. Our solution artifact provides a lightweight AI application that is adoptable for other problem spaces, enabling an easier knowledge transfer from research to SMEs. By combining Cloud Computing with Machine Learning, the artifact identifies present and emerging damages of physical objects. It provides a virtual representation of the object's state and empowers a meaningful visualization. 2021-01-01T00:00:00Z http://dl.gi.de/handle/20.500.12116/37767 Chameleon: A Semi-AutoML framework targeting quick and scalable development and deployment of production-ready ML systems for SMEs Otterbach, Johannes; Wollmann, Thomas Developing, scaling, and deploying modern Machine Learning solutions remains challenging for small- and middle-sized enterprises (SMEs). This is due to a high entry barrier of building and maintaining a dedicated IT team as well as the difficulties of real-world data (RWD) compared to standard benchmark data. To address this challenge, we discuss the implementation and concepts of Chameleon, a semi-AutoML framework. The goal of Chameleon is fast and scalable development and deployment of production-ready machine learning systems into the workflow of SMEs. We first discuss the RWD challenges faced by SMEs. After, we outline the central part of the framework which is a model and loss-function zoo with RWD-relevant defaults. Subsequently, we present how one can use a templatable framework in order to automate the experiment iteration cycle, as well as close the gap between development and deployment. Finally, we touch on our testing framework component allowing us to investigate common model failure modes and support best practices of model deployment governance. 2021-01-01T00:00:00Z http://dl.gi.de/handle/20.500.12116/37768 Energieeffizientes Kaltstartverhalten spanender Werkzeugmaschinen Walz, Deborah; Wächter, Andreas; Tomov, Stefan; Heimbach, Konrad; Weigold, Matthias Die Kompensation thermischer Einflüsse und daraus resultierender geometrischer Verlagerungen spielt eine bedeutende Rolle bei der Gewährleistung einer hohen Bearbeitungsqualität von Werkstücken in Zerspanungsprozessen. Übliche Vorgehensweisen zur Reduktion thermischer Verlagerungen während der Produktion gehen mit einem erheblichen Energiebedarf einher oder modellieren die komplexen Zusammenhänge thermischer Einflüsse nur ungenügend. Methoden des Maschinellen Lernens stellen einen vielversprechenden Ansatz zur Modellierung dar. Es wird eine Lösung angestrebt, die aufwandsarm auf Produktionsmaschinen ähnlicher Bauart übertragen werden kann. Derzeit ist ungeklärt, ob eine explizite oder implizite Modellierung der zeitlich multivarianten Daten eine ufriedenstellende Lösung bietet. Als besonders herausfordernd stellt sich die Verfügbarkeit von ausreichend vielen Datenbeispielen zur Modellierung der relevanten Größen dar. 2021-01-01T00:00:00Z http://dl.gi.de/handle/20.500.12116/37765 Blueprint for a Production-Ready Information Retrieval System based on Multi-Modal Embeddings Ebert, André; Apel, Anika; Chodyko, Piotr; Hiroyasu, Kyle; Ismali, Festina; Koo, Hyein; Kronburger, Julia; Pesch, Robert Deep Learning models for mapping documents from different domains, e.g., text, images, and audio, into a common vector space, enable a seamless information retrieval between the different domains and, thus, significantly improve the user experience of many expert tools. Despite various models for multi-modal mappings presented in scientific literature, the implementation and integration remain a challenge within the industry, especially for small or medium-sized companies. Reasons are, that developing such retrieval systems for production use-cases is a non-trivial task, requiring scalable, reliable, and cost-efficient infrastructure, services as well as adequate Deep Learning models. We present a generic and flexible blueprint architecture, targeting the development of a production-ready image-text retrieval search system using Kubernetes, MLflow, Elasticsearch, and integrating state-of-the-art Deep Learning models. 2021-01-01T00:00:00Z