P344 - 44. GIL-Jahrestagung 2024 - Fokus: Biodiversität fördern durch digitale Landwirtschaft

Permanent URI for this collectionhttps://dl.gi.de/handle/20.500.12116/43863

Authors with most Documents  

Browse

Search Results

1 - 5 of 5
  • Conference Paper
    Adaptive real-time crop row detection through enhancing a traditional computer vision approach
    (Gesellschaft für Informatik e.V., 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.
  • Conference Paper
    Exploring AI for interpolation of combine harvester yield data
    (Gesellschaft für Informatik e.V., 2024) Johannsen, Lucas; Ramm, Sebastian; Reckleben, Yves; Doerfel, Stephan
    In the wake of eco-schemes introduced by the EU's Common Agricultural Policy, this study evaluates AI-based interpolation methods for generating yield maps as one component of a decision support system, aiding farmers in eco-scheme implementation. The research contrasts ordinary Kriging (OK) with AI techniques – Random Forest (RF) enhanced with spatial fea-tures (RFsp), covariates (RFspco) and DeepKriging (DK), utilizing combine harvester yield data. Performance metrics show AI, especially RF variants, surpassing OK. For a 0.7 split, R² were 0.6 (OK), 0.77 (RF), 0.81 (RFsp), 0.78 (DK); MSE were 0.6 (OK), 0.34 (RF), 0.28 (RFsp), 0.32 (DK). Spatial features boosted accuracy, while incorporating Terrain Models had no rele-vant impact on the results. These findings are crucial for an automated, accurate decision support system, facilitating eco-scheme adoption for farmers. The efficiency of AI methods underscores their potential in promoting sustainable, informed agricultural practices.
  • Conference Paper
    A crowdsensing-based smartphone app for optimal food storage and real-time best-before dates
    (Gesellschaft für Informatik e.V., 2024) Senge, Julia; Mielinger, Ellen; Wendt, Marie Catherine; Weinrich, Ramona; Krupitzer, Christian
    Private households are responsible for 59% of Germany’s 11 million tons of food waste. Consumers’ behavior significantly contributes to food waste, prompting our concept to develop a smartphone application aimed at diminishing uncertainties about food expiration and safety. Utilizing a Design Science approach, we developed a prototype for a smartphone app, integrating novel functionalities to minimize food waste at the consumer household level. We analyzed existing market applications and, as a result, introduced the Freshlimeter, a unique feature that estimates the real-time best-before date within our app using feedback from consumers. We also highlight the potential for innovative app features, such as integrating a chatbot with image recognition capabilities to enable freshness assessments, especially for unpackaged or opened food.
  • Conference Paper
    Learning from hyperspectral remote sensing data for machine learning algorithms in earth science
    (Gesellschaft für Informatik e.V., 2024) Christoph Jörges, Sandra Dotzler
    Machine Learning in Earth sciences heavily depends on sufficient training data for proper generalization. Since in-situ ground truth data is rarely available and cost-intensive to obtain, this study presents a new approach of deriving training data from hyperspectral remote sensing satellites by physical spectral signatures to use them for data-driven models with operationally available multispectral data. Examples include monitoring of crop rotation, winter greening, soil organic matter, and detection of plastic covered greenhouses (PCGs).
  • Conference Paper
    A comparative study of RGB and multispectral imaging for weed detection in precision agriculture
    (Gesellschaft für Informatik e.V., 2024) Benedikt Fischer, Pascal Gauweiler
    Precision agriculture and specifically mechanical weed control systems have the potential to positively impact our environment by reducing the use of herbicides. In recent years, multispectral cameras have become more and more accessible, which raises the question whether the additional costs of such cameras are worth the potential benefits. In this study, we recorded and annotated a multispectral instance segmentation dataset for sugar beet crop and weed detection. We trained Mask-RCNN models on the RGB and multispectral data in a transfer learning approach and extensively evaluated and compared the results for different scenarios. We found that the multispectral data can improve the weed detection performance significantly in many cases.
Load citations