Logo des Repositoriums

Auflistung nach Autor:in "Hussaini, Mortesa"

1 - 4 von 4
  • Konferenzbeitrag
    Adaptive real-time crop row detection through enhancing a traditional computer vision approach
    (44. GIL - Jahrestagung, Biodiversität fördern durch digitale Landwirtschaft, 2024) Hussaini, Mortesa; Voigt, Max; Stein, Anthony
    Crop row detection is important to enable precise management of fields and optimize the use of resources such as fertilizers and water. Autonomous machines need an effective but also robust real-time row detection system to be able to adapt to different field conditions. In this paper, we present an enhanced crop row detection approach which integrates traditional computer vision methods with further techniques such as k-means clustering or probabilistic Hough transformation. The resulting hybrid method allows for efficient and robust detection of straight and curved crop rows in image and video material. We validate our approach empirically on the crop row benchmark dataset (CRBD) and compare it with other state-of-the-art approaches. Furthermore, we demonstrate that our approach is designed to be adaptive and thus becomes straightforwardly transferable to other experimental setups. To corroborate that, we report on results when our approach is validated on representative corner cases which have been collected in the scope of a research project. Observations and current limitations of our approach are discussed along with possible solutions to overcome them in future work.
  • Konferenzbeitrag
    Federated Learning in Agriculture: Potential and Challenges
    (INFORMATIK 2023 - Designing Futures: Zukünfte gestalten, 2023) Hussaini, Mortesa; Stein, Anthony
    Federated learning is an emerging technique in machine learning that allows multiple parties to collaboratively train models without sharing raw data. It has been applied in various fields such as healthcare, finance, and transportation. In this paper, we cast light on the potential of federated learning in the highly relevant social-ecological domain of agriculture, a field in which digitization is becoming increasingly prevalent. We briefly introduce the collaborative learning concept of federated learning and briefly consider its postulated benefits and open challenges. The potential of federated learning to overcome concerns against digital technology in agriculture, e.g., data privacy and sovereignty or initial investment and operating costs, is then discussed. We also identify system requirements and stress the necessity of appropriate IT-ecosystems and touch upon specific requirements which can enable federated learning to ensure both a data-sovereign and efficient information and knowledge exchange among multiple parties. Based on that, agricultural use cases where federated learning can unfold its potential by not only improving the quality of machine learning models, but also by alleviating overarching adoption barriers, will be exemplary delineated.
  • Konferenzbeitrag
    Multi-level Personalised Federated Learning: A concept for data sovereign machine learning in digital agriculture
    (INFORMATIK 2024, 2024) Hussaini, Mortesa; Stein, Anthony
    Securing food supply and protecting the environment at the same time are two of the major challenges modern agriculture is facing. Research into Artificial Intelligence has seen an enormous increase in interest and potential fields of application. In agriculture, farmers can benefit enormously from this technology. However, the diverse potential is also contrasted by a number of socio-economic concerns and challenges. This aspect needs to be addressed in order to foster a trustworthy use and proliferation of AI applications in socio-technical systems, such as digital agriculture. Federated Learning is a ML paradigm that constitutes a promising solution to this challenge. However, in agricultural use cases exhibiting high degrees of data heterogeneity, client-drift can occur and needs to be addressed. In this concept paper, we propose the advancement of Personalised Federated Learning with multiple levels of sub-federations as a potential solution.
  • Konferenzbeitrag
    Nachhaltige Landwirtschaft mittels Künstlicher Intelligenz – ein plattformbasierter Ansatz für Forschung und Industrie
    (43. GIL-Jahrestagung, Resiliente Agri-Food-Systeme, 2023) Bosse, Sebastian; Berns, Karsten; Bosch, Johannes; Dörr, Jörg; Eichhorn, Frederick Charles; Eisert, Peter; Fischer, Christoph; Gassen, Eike; Gerstenberger, Michael; Gerighausen, Heike; Heil, Jonathan; Hilsmann, Anna; Hirth, Jochen; Huber, Christopher; Hussaini, Mortesa; Kasparick, Martin; Kloke, Peter; Krause-Edler, Hartmut; Mackle, Lukas; Magnusson, Jannes; Möhrle, Felix; Möller, Markus; Pickel, Peter; Rautenberg, Clemens; Schotten, Hans Dieter; Stanczak, Slawomir; Thiele, Lars; Ücdemir, Henrik; Wania, Annett; Stein, Anthony
    Digitale Technologien gelten als möglicher Schlüssel zur Verknüpfung von Nachhaltigkeit, Klimaanpassung und wirtschaftlicher Effizienz in der Pflanzenproduktion. Die Heterogenität und Dezentralität des landwirtschaftlichen Systems stellt besondere Anforderungen an den Entwurf datengetriebener Lösungen: Daten entstehen lokal in landwirtschaftlichen Betrieben unterschiedlicher Größe; ihre Erhebung und Auswertung erfolgt meist multimodal, dezentral und durch Dritte; landwirtschaftliche Stakeholder stellen als Dateneigentümer hohe Ansprüche an die Datensouveränität. Das Forschungsprojekt „Nachhaltige Landwirtschaft mittels Künstlicher Intelligenz“ (NaLamKI) entwickelt einen plattformbasierten Ansatz, um diese Anforderungen zu adressieren und setzt hierzu auf Cloud-Edge Services zur 1) Erhebung divers strukturierter landwirtschaftlicher Daten, 2) KI-gestützte Fusion und Auswertung dieser Daten sowie 3) nutzerorientierte Haltung und Bereitstellung der erzeugten Datenprodukte in einem digitalen Farm-Twin unter Wahrung der Datensouveränität, Schaffung von (Daten-)Interoperabilität sowie GAIA-X-Konformität. Dieser Beitrag leitet die Notwendigkeit dieses Forschungsansatzes her, erläutert dessen zugrunde liegende Konzepte und diskutiert wissenschaftliche Ansatzpunkte und Ergebnisse sowie offene Herausforderungen und Chancen dieses integrierten Ansatzes.