Browsing by Person "Allmendinger, Alicia"

Now showing 1 - 3 of 3

Agronomic and technical evaluation of herbicide spot spraying in maize based on high-resolution aerial weed maps - an on-farm trial
(2024) Allmendinger, Alicia; Spaeth, Michael; Saile, Marcus; Peteinatos, Gerassimos G.; Gerhards, Roland; Allmendinger, Alicia; Department of Weed Science, Institute for Phytomedicine, University of Hohenheim, 70599 Stuttgart, Germany; (A.A.);; Spaeth, Michael; Department of Weed Science, Institute for Phytomedicine, University of Hohenheim, 70599 Stuttgart, Germany; (A.A.);; Saile, Marcus; Department of Weed Science, Institute for Phytomedicine, University of Hohenheim, 70599 Stuttgart, Germany; (A.A.);; Peteinatos, Gerassimos G.; ELGO-DIMITRA, Leof Dimokratias 61, Agii Anargiri, 135 61 Athens, Greece;; Gerhards, Roland; Department of Weed Science, Institute for Phytomedicine, University of Hohenheim, 70599 Stuttgart, Germany; (A.A.);; Rossi, Vittorio
Spot spraying can significantly reduce herbicide use while maintaining equal weed control efficacy as a broadcast application of herbicides. Several online spot-spraying systems have been developed, with sensors mounted on the sprayer or by recording the RTK-GNSS position of each crop seed. In this study, spot spraying was realized offline based on georeferenced unmanned aerial vehicle (UAV) images with high spatial resolution. Studies were conducted in four maize fields in Southwestern Germany in 2023. A randomized complete block design was used with seven treatments containing broadcast and spot applications of pre-emergence and post-emergence herbicides. Post-emergence herbicides were applied at 2–4-leaf and at 6–8-leaf stages of maize. Weed and crop density, weed control efficacy (WCE), crop losses, accuracy of weed classification in UAV images, herbicide savings and maize yield were measured and analyzed. On average, 94% of all weed plants were correctly identified in the UAV images with the automatic classifier. Spot-spraying achieved up to 86% WCE, which was equal to the broadcast herbicide treatment. Early spot spraying saved 47% of herbicides compared to the broadcast herbicide application. Maize yields in the spot-spraying plots were equal to the broadcast herbicide application plots. This study demonstrates that spot-spraying based on UAV weed maps is feasible and provides a significant reduction in herbicide use.
Herbicide reduction through Convolutional Neural Network (CNN)-based technologies for advanced weed control
(2024) Allmendinger, Alicia; Gerhards, Roland
Weeds are plants that can grow in agricultural fields and can compete with cultivated crops for essential resources, including light, water and nutrients. Such competition can result in yield losses, thereby necessitating the implementation of weed control measures by farmers. Traditional weed control is performed by the use of herbicides, which in previous years were applied in a uniform manner across the entire field. In recent years, there has been a growing awareness of the detrimental impact of herbicides on the environment, as well as on human health. Furthermore, the impact on the crop, biodiversity in agriculture and fauna was examined. Given that herbicides have been found in both groundwater and the food chain, there is an urgent need to develop alternative methods for site-specific weed control, rather than a uniform application. In addition to the complete avoidance of herbicides through mechanical weed control, such as in-row hoeing, methods including site-specific herbicide application and spot-spraying, represent a promising approach for consideration. In the context of a site-specific weed control the use of herbicides is confined to areas where weeds are actually present. Additionally, the use of a selective herbicide can be employed, which is targeted to specific weed species. Furthermore, the application of species is possible above a specific threshold. Site-specific weed control includes methods like spot-spraying, where only single weeds are treated, and also mechanical solutions such as in-row hoeing, which can also control weeds intra row. However, reliable detection and identification of the plants or plant species is essential to implement any of these methods effectively. The process of recognition in digital images is currently often facilitated by convolutional neural networks (CNNs). These systems are either used as online or offline systems. In online systems images are gathered with f.e. a camera mounted to the front of a tractor and plant species are identified directly on a tractor´s onboard computer by an image classifier. This process allows a real-time application of herbicides during a single pass. In offline systems, the field is scanned in advance by for example unmanned aerial vehicles (UAVs), which capture georeferenced images of the field. These images are analysed in the next step by an image classifier, which generates a weed control map, that is later uploaded to the tractor's terminal and an application of herbicides based on RTK-GNNS can be performed. This dissertation presents a summary of the current research in this field and offers an empirical test of an offline system in a practical agricultural setting. Moreover, multiple CNNs were trained using weed and crop images, and their efficacy for real-time detection was evaluated. The first article presents an overview of the current research in the domain of site-specific herbicide application. The article provides a summary of the existing technology and situates it within the context of current trends. Furthermore, a CNN-based modular spot sprayer is presented, which has been designed to facilitate communication between the tractor and the mounted implement via an ISOBUS connection. The second article addresses the utilisation of unmanned aerial vehicles (UAVs) for the generation of weed control maps in maize. Field trials were conducted on four fields at two locations in 2023 as part of the study. The efficacy of pre-emergence and post-emergence herbicides was evaluated and the post-emergence herbicide was applied in both broadcast and spot application. The timing of the post-emergence herbicide application was varied, with the first application occurring at the two-to-four-leaf stage and the second at the six-to-eight-leaf stage of the maize. The image classifier utilized for the weed control maps exhibited a detection accuracy of 94%. Weed control efficacy in spot spraying was 86%, comparable with the efficacy of the broadcast application. Using weed control maps resulted in a reduction of herbicide by up to 47% without any negative impact on the yield. This approach has the potential to reduce operational costs and the negative impact of herbicides on the environment. The third article presents the use of different CNNs for the differentiation between diverse plant species and the differentiation between crops, monocotyledonous weeds and dicotyledonous weeds. Therefore, all models of the networks YOLOv8, YOLOv9, YOLOv10 and RT-DETR that were available at that time were trained on different central processing units (CPUs) and graphic processing units (GPUs). The results demonstrated that a differentiation between individual weed species is feasible with a mean Average Precision (mAP) of up to 43.82 % at an Intersection over Union (IoU) threshold of 0.5 to 0.95. Nevertheless, greater accuracy can be attained by distinguishing only between crops, monocot and dicot weeds with mAP50-95 scores of up to 47.26 %. Moreover, it is evident that there are considerable fluctuations in inference time across the various models, with detection times ranging from 8.46 ms for YOLOv8n on the NVIDIA GeForce RTX 3090 GPU to 120.44 ms on the AMD Ryzen 9 5950X 16-Core CPU for species wise object detection. These findings underscore that GPU´s, with their faster processing times, are most suited for real-time application. The studies conducted show that precise distinguishing between plant species, and between weeds and crops, is achievable with precision scores exceeding 80 %. Nevertheless, for site-specific herbicide application or spot spraying this level of accuracy may be sufficient. In other applications, such as in-row hoeing, an accuracy of 80% may be insufficient, as this might cause great crop losses. Nevertheless, the implementation of CNN-based technologies, particularly the emerging RT-DETR, has the potential to significantly reduce the amount of herbicides used, thereby achieving the objectives of the EU Green Deal. Furthermore, these techniques allow for the implementation as online systems where suitable, as well as offline systems when more appropriate. However, for online systems it is essential to verify that the on-board computer of the tractor provides the necessary computational resources to support real-time applications. Additionally, for practical deployment on farms, the spraying technology must be available, with the compatible nozzles, that ensure for example accurate and effective spot spraying. Overall, spot-spraying can only be conducted with post-emergence herbicides; however, it would be beneficial reducing or even avoiding pre-emergence herbicides in order to achieve further savings, as these necessitate a broadcast application. When all of these optimisations are considered, it becomes evident that the objectives of the EU Green Deal are being met without compromising agricultural production yield.
Precision chemical weed management strategies: A review and a design of a new CNN-based modular spot sprayer
(2022) Allmendinger, Alicia; Spaeth, Michael; Saile, Marcus; Peteinatos, Gerassimos G.; Gerhards, Roland
Site-specific weed control offers a great potential for herbicide savings in agricultural crops without causing yield losses and additional weed management costs in the following years. Therefore, precision weed management is an efficient tool to meet the EU targets for pesticide reduction. This review summarizes different commercial technologies and prototypes for precision patch spraying and spot spraying. All the presented technologies have in common that they consist of three essential parts. (1) Sensors and classifiers for weed/crop detection, (2) Decision algorithms to decide whether weed control is needed and to determine a suitable type and rate of herbicide. Usually, decision algorithms are installed on a controller and (3) a precise sprayer with boom section control or single nozzle control. One point that differs between some of the techniques is the way the decision algorithms classify. They are based on different approaches. Green vegetation can be differentiated from soil and crop residues based on spectral information in the visible and near-infrared wavebands (“Green on Brown”). Those sensors can be applied for real-time on/off control of single nozzles to control weeds before sowing after conservation tillage and in the inter-row area of crops. More sophisticated imaging algorithms are used to classify weeds in crops (“Green on Green”). This paper will focus on Convolutional Neural Networks (CNN) for plant species identification. Alternatively, the position of each crop can be recorded during sowing/planting and afterward herbicides can be targeted to single weeds or larger patches of weeds if the economic weed threshold is exceeded. With a standardized protocol of data communication between sensor, controller and sprayer, the user can combine different sensors with different sprayers. In this review, an ISOBUS communication protocol is presented for a spot sprayer. Precision chemical weed control can be realized with tractor-mounted sprayers and autonomous robots. Commercial systems for both classes will be introduced and their economic and environmental benefits and limitations will be highlighted. Farmers ask for robust systems with less need for maintenance and flexible application in different crops.