Institut für Agrartechnik

Permanent URI for this collectionhttps://hohpublica.uni-hohenheim.de/handle/123456789/19

Browse

Now showing 1 - 13 of 13

Assessing the capability of YOLO- and transformer-based object detectors for real-time weed detection
(2025) Allmendinger, Alicia; Saltık, Ahmet Oğuz; Peteinatos, Gerassimos G.; Stein, Anthony; Gerhards, Roland
Assessing the capability of YOLO- and transformer-based object detectors for real-time weed detection
(2025) Allmendinger, Alicia; Saltık, Ahmet Oğuz; Peteinatos, Gerassimos G.; Stein, Anthony; Gerhards, Roland
Spot spraying represents an efficient and sustainable method for reducing herbicide use in agriculture. Reliable differentiation between crops and weeds, including species-level classification, is essential for real-time application. This study compares state-of-the-art object detection models-YOLOv8, YOLOv9, YOLOv10, and RT-DETR-using 5611 images from 16 plant species. Two datasets were created, dataset 1 with training all 16 species individually and dataset 2 with grouping weeds into monocotyledonous weeds, dicotyledonous weeds, and three chosen crops. Results indicate that all models perform similarly, but YOLOv9s and YOLOv9e, exhibit strong recall (66.58 % and 72.36 %) and mAP50 (73.52 % and 79.86 %), and mAP50-95 (43.82 % and 47.00 %) in dataset 2. RT-DETR-l, excels in precision reaching 82.44 % (dataset 1) and 81.46 % (dataset 2) making it ideal for minimizing false positives. In dataset 2, YOLOv9c attains a precision of 84.76% for dicots and 78.22% recall for Zea mays L.. Inference times highlight smaller YOLO models (YOLOv8n, YOLOv9t, and YOLOv10n) as the fastest, reaching 7.64 ms (dataset 1) on an NVIDIA GeForce RTX 4090 GPU, with CPU inference times increasing significantly. These findings emphasize the trade-off between model size, accuracy, and hardware suitability for real-time agricultural applications.
Computational sizing of solar powered peanut oil extraction in Senegal using a synthetic load profile
(2024) Bonzi, Joévin Wiomou; Romuli, Sebastian; Diouf, Djicknoum; Piriou, Bruno; Meissner, Klaus; Müller, Joachim
This paper presents an approach for sizing a hybrid photovoltaic system for a small-scale peanut oil processing company (Yaye Aissatou, Passy) in rural Senegal using a synthetic load profile. In this study, a predictive model of the electrical load of a service-based plant oil processing company was developed through a diagnosis, to evaluate the extraction process. The mass and energy balance were measured, and the process was implemented into MATLAB Simulink. The simulated load profile was implemented in HOMER Pro and the characteristics of the most profitable hybrid systems were identified. The results showed that the lowest net present cost over 25 years was found with a PV/battery/grid-system with 18.6 kWp solar panels, 16 kWh of storage, and an initial investment of 20,019 €. Compared to a grid-only scenario, this solution reduces the net present cost from an initial 72,163 € to 31,603 €, the operating cost from 3675 € per year to 590 € per year, and the cost of energy from 0.29 to 0.13 €/kWh. The renewable fraction of the proposed system is 90.0 % while the expected payback period is 6.2 years. The study demonstrates the economic feasibility of using solar energy for plant oil processing.
Disc mower versus bar mower: Evaluation of the direct effects of two common mowing techniques on the grassland arthropod fauna
(2025) von Berg, Lea; Frank, Jonas; Betz, Oliver; Steidle, Johannes L. M.; Böttinger, Stefan; Sann, Manuela
1. In Central Europe, species‐rich grasslands are threatened by intensive agriculture with frequent mowing, contributing to the reduction of arthropods such as insects and spiders. However, comprehensive and standardised studies on the direct effects of the two most agriculturally relevant mowing techniques, e.g., double‐blade bar mower versus disc mower, are lacking. 2. In a 2‐year experiment, we have investigated the direct effect of mowing on eight abundant arthropod groups in grassland, covering two seasonal mowing events in both years, using a randomised block design. We compared (a) an unmown control, (b) a double‐blade bar mower and (c) a disc mower. 3. For most of the taxonomic groups studied, a significantly lower number of individuals was found in the experimental plots immediately after mowing, regardless of the mowing technique, compared to an unmown control. This was not the case for Orthoptera and Coleoptera, which did not show a significant reduction in the number of individuals for both mowing techniques (Orthoptera) or only for the double‐blade bar mower (Coleoptera). 4. Between both mowing techniques, no significant differences were found for all taxonomic groups investigated. 5. Synthesis and applications: Our findings suggest that mowing in general has a negative impact on abundant arthropod groups in grassland, regardless of the method used. Tractor‐driven double‐blade bar mowers do not seem to be a truly insect‐friendly alternative to a conventional disc mower. Other factors such as cutting height and mowing regimes should be seriously considered to protect spiders and insects from the negative effects of mowing. In addition, we strongly recommend the maintenance of unmown refugia. Insects and spiders that are spared by mowing can take refuge in these unmown areas to avoid subsequent harvesting and thermally unfavourable conditions that arise on mown areas. Further, unmown refugia are basic habitat structures for a subsequent recolonisation of mown areas once the flora has recovered.
Effects of different ground segmentation methods on the accuracy of UAV-based canopy volume measurements
(2024) Han, Leng; Wang, Zhichong; He, Miao; He, Xiongkui
The nonuniform distribution of fruit tree canopies in space poses a challenge for precision management. In recent years, with the development of Structure from Motion (SFM) technology, unmanned aerial vehicle (UAV) remote sensing has been widely used to measure canopy features in orchards to balance efficiency and accuracy. A pipeline of canopy volume measurement based on UAV remote sensing was developed, in which RGB and digital surface model (DSM) orthophotos were constructed from captured RGB images, and then the canopy was segmented using U-Net, OTSU, and RANSAC methods, and the volume was calculated. The accuracy of the segmentation and the canopy volume measurement were compared. The results show that the U-Net trained with RGB and DSM achieves the best accuracy in the segmentation task, with mean intersection of concatenation (MIoU) of 84.75% and mean pixel accuracy (MPA) of 92.58%. However, in the canopy volume estimation task, the U-Net trained with DSM only achieved the best accuracy with Root mean square error (RMSE) of 0.410 m 3 , relative root mean square error (rRMSE) of 6.40%, and mean absolute percentage error (MAPE) of 4.74%. The deep learning-based segmentation method achieved higher accuracy in both the segmentation task and the canopy volume measurement task. For canopy volumes up to 7.50 m 3 , OTSU and RANSAC achieve an RMSE of 0.521 m 3 and 0.580 m 3 , respectively. Therefore, in the case of manually labeled datasets, the use of U-Net to segment the canopy region can achieve higher accuracy of canopy volume measurement. If it is difficult to cover the cost of data labeling, ground segmentation using partitioned OTSU can yield more accurate canopy volumes than RANSAC.
Experimental analysis and CFD-based modeling of grain bulk drying dynamics
(2021) Ramaj, Iris; Schock, Steffen; Müller, Joachim
Drying is of great importance in the postharvest processing of agricultural commodities. It refers to the removal of the surplus moisture responsible for biochemical, microbiological, and other moisture-related deteriorative reactions, thereby ensuring quality preservation. However, drying is an intricate process comprising simultaneous heat and moisture transfers, which depends on product and drying air conditions. Therefore, drying practices are oftentimes misused, resulting in serious degradation of product quality. For this reason, modeling can be used to provide a deeper understanding of air-product interactions and to gain insights into drying process. Thus, this study focused on developing a CFD systematic approach to model the drying dynamics of wheat bulk (Pionier A, DSV AG) under controlled conditions. Within the model framework, a porous medium approach with tailored user-defined functions was utilized to represent the grain bulk characteristics. The drying experiments were performed using a high-precision and automated through-flow laboratory dryer. A coherent set of drying air temperatures T = 10 - 50°C, relative humidity RH = 20 - 60% and airflow velocity v = 0.15 - 1.00 ms-1 were employed for model validation. Afterwards, the validated computational model was used to predict the drying performance at T = 40°C, RH = 40% and v = 0.15 ms-1, where the simulated temperature and moisture content agreed very well with the experimental results (R2 ≥ 0.98 and MAPE ≤ 14.93%). The proposed model proved to be an efficient tool capable of simulating temperature and moisture dynamics inside the grain bulk with high spatial and temporal resolution, providing rapid and in-depth information compared to laborious physical experiments. In conclusion, the CFD-based approach has demonstrated a great potential for simulating drying processes. Its capabilities should be further assessed across various drying technologies, operating conditions, and agricultural commodities.
Food informatics - Review of the current state-of-the-art, revised definition, and classification into the research landscape
(2021) Krupitzer, Christian; Stein, Anthony
Background: The increasing population of humans, changing food consumption behavior, as well as the recent developments in the awareness for food sustainability, lead to new challenges for the production of food. Advances in the Internet of Things (IoT) and Artificial Intelligence (AI) technology, including Machine Learning and data analytics, might help to account for these challenges. Scope and Approach: Several research perspectives, among them Precision Agriculture, Industrial IoT, Internet of Food, or Smart Health, already provide new opportunities through digitalization. In this paper, we review the current state-of-the-art of the mentioned concepts. An additional concept is Food Informatics, which so far is mostly recognized as a mainly data-driven approach to support the production of food. In this review paper, we propose and discuss a new perspective for the concept of Food Informatics as a supportive discipline that subsumes the incorporation of information technology, mainly IoT and AI, in order to support the variety of aspects tangent to the food production process and delineate it from other, existing research streams in the domain. Key Findings and Conclusions: Many different concepts related to the digitalization in food science overlap. Further, Food Informatics is vaguely defined. In this paper, we provide a clear definition of Food Informatics and delineate it from related concepts. We corroborate our new perspective on Food Informatics by presenting several case studies about how it can support the food production as well as the intermediate steps until its consumption, and further describe its integration with related concepts.
Improvement of the drying performance of pre-cooked beans (Phaseolus vulgaris) through ultrasonic-assisted hulling
(2022) Ramaj, Iris; Schock, Steffen; Ayetigbo, Oluwatoyin; Ntwali, Janvier; Müller, Joachim
Beans are among the most versatile and widely consumed staple foods worldwide. They are highly nutritious and contain high levels of dietary fibers, complex carbohydrates, proteins, essential vitamins, and minerals that are indispensable to human wellbeing. Due to their given importance, the development of processing methods for hard-to-cook beans for the preparation of instant end-products is of great interest, especially in developing countries. Thus, this study focused on investigating the influence of ultrasonic-assisted dehulling on the drying behaviour of pre-cooked beans as a viable alternative to the present drying approaches. Red kidney beans (Phaseolus vulgaris), unhulled (UHB) and dehulled via ultrasonication (HB/UT), were used for the experimental analysis. The cooking time of beans was determined based on sensory evaluation, with 50 and 25 min proving to be optimal for UHB and HB/UT, respectively. Afterwards, the pre-cooked samples were dried in a high-precision through-flow laboratory dryer (HPD-TF3+) at 30, 50, and 70° C with an air velocity of 0.20 ms-1 and specific humidity of 10 g kg-1. Results revealed a faster moisture transfer of the HB/UT beans compared to UHB beans at p < 0.05, which was attributed to the lower resistance to moisture diffusion induced by the hull removal. Henceforth, a reduction of drying time up to 73.3% was ascertained experimentally. A generalised semi-empirical model was developed from the analysis of the drying data, which was capable of predicting the drying behaviour of beans with R2 ≥ 0.990 and MAPE ≤ 10.0%. In terms of colour, UHB and HB/UT beans differed significantly at p < 0.05 for redness a*, yellowness b*, hue angle H*, and chroma c* across all drying conditions, while no significant differences were observed for luminosity L*. Microstructural analysis revealed comparable structures after drying at 30 and 50° C, with beans exhibiting an intact cellular structure. Temperatures of 70° C, on the other hand, degraded the cellular integrity of beans by breaking down the cell wall boundaries, especially in HB/UT beans. In conclusion, ultrasonic-assisted hulling has demonstrated a great potential for improving the drying performance of beans, thereby making it a viable alternative for practical applications.
Influence of self-compaction on the airflow resistance of grain bulks
(2020) Ramaj, Iris; Schock, Steffen; Karaj, Shkelqim; Müller, Joachim
Aeration practices have been widely employed to force conditioned air through in-storage grain bulks to guarantee quality preservation and safe storage. Despite the attention given in the last decades as a principal grain management technique, many obstacles have been encountered in reducing the deteriorative effects of high moisture content and temperature throughout the in-storage bulk. This was attributed to the misestimation of the resistance of stored grains to the airflow, which led to deficiencies in bulk aeration. This resistance is complex and strongly dependent on airflow and grain properties. In this study, the airflow resistance of wheat grains (Pioneer A DSV AG, 12.37 % w.b moisture content) during storage was investigated. A cylindrical, stationary bed (0.5 m diameter and 3.6 m height) was used as an experimental basis. A coherent set of airflow velocities ranging from 0.01 to 0.15 m·s-1 and storage times ranging from 1 to 236 h at four grain depth levels were applied accordingly. The relationship between pressure drop and velocity was assessed experimentally and modeled theoretically with an overall goodness of fit of R2=0.99, RMSE=25.7, and MAPE=10.4%. Results demonstrated an increase of the airflow resistance throughout the depth of the grain bulk and storage time. This behavior was ascribed to the self-compaction of the bulk material arising from the burden pressures imposed by the dead weight of the bulk. The self-compaction decreased the porosity significantly, increased the bulk density, enlarged the airflow resistance and consequently, considerably increased the pressure drop. Hence, extra power supplies for aeration are prerequisites to overcome the resistance caused by self-compaction. The spatial and temporal effects of self-compaction in stored grain bulks should be accommodated in the design and analysis of aeration systems.
Investigation of the spatial and temporal variations of weather conditions in a mesoscale vineyard
(2021) Pfisterer, Philipp; Schock, Steffen; Ramaj, Iris; Müller, Joachim
Climatological conditions and weather variability have a momentous impact on viticulture and vineyard management and can be detrimental for grapevine growth and its yield. Humid weather conditions contribute to the spread of fungal pathogens and diseases, which afterwards degrade the quality of the grapevine and risk the longevity of orchards and vineyards in the tropical and subtropical regions. Therefore, it is critical to monitor the spatial and temporal variations of weather conditions in the vineyards. Despite numerous sensor systems developed in academia and industry to address this problem, a scalable and dense sensor system that guarantees low maintenance, fast and reliable data acquisition is still lacking. Thus, in this study, a low-cost wireless networked system was developed for real-time monitoring of weather parameters namely, temperature, relative humidity and dew point temperature. A capacitive-type sensor SHT31 integrated into an STM32L0 microcontroller was employed as a measuring unit. Data transmittance was empowered via a functional radio network. The sensor housings were designed and manufactured in-house via a 3D printer. The accuracy of data readings was validated by a climatic test chamber CTS-20/1000 under a wide-ranging set of temperatures and relative humidities. As an evasive experimental site, a mesoscale 30-ha vineyard located in Hessigheim, Germany was used to test the monitoring system. A number of 30 sensors were installed irregularly in this area. For the graphical analysis, data collected during the summer and winter periods were compared. From the results, substantial differences in temperature were observed between vineyard sites at p ≤ 0.05. The spatial temperature gradients altered up to 8°C, which was mainly attributed to the heterogeneous and steeply sloping terrain of the vineyard. These gradients increased over the summer and decreased during the winter. This behaviour was accredited to the diurnal solar orientation, shaded conditions as well as wind direction imposed by a bend in the river. Likewise, significant differences were observed for dew point and relative humidity. In conclusion, the developed network system demonstrated a high capability to track the variability of weather conditions and should be used as a tool for the prediction of infection hotpots in vineyards.
Multi-crop early detection of spider mite damage using hyperspectral data and XGBoost
(2026) Mandrapa, Boris; Spohrer, Klaus; Wuttke, Dominik; Ruttensperger, Ute; Dieckhoff, Christine; Müller, Joachim
The two-spotted spider mite is a globally significant pest affecting over 150 crop species, including cucumbers and strawberries. Its feeding activity leads to chlorophyll degradation and physiological changes in leaf tissue, which alter spectral reflectance properties and enable image-based detection. In this study, hyperspectral imaging (HSI) under controlled conditions was used to classify healthy and spider mite-infested leaves of cucumber and strawberry plants, including asymptomatic infested leaves. Spectral data were analyzed and classified with three supervised machine learning algorithms built on extreme gradient boosting (XGBoost) models. The study had three objectives: (1) to assess the ability of XGBoost to classify multiple infestation states, (2) to evaluate model performance with a reduced set of effective wavelengths, and (3) to determine whether infestation across both crops can be classified using a single, merged model. Using all wavelengths, results showed that classification accuracy was 93 % for cucumber leaves, 84 % for strawberry leaves, and 87 % when combined. With five most effective wavelengths, classification accuracy reached 70 % for cucumber leaves, 65 % for strawberry leaves, and 65 % for cucumber and strawberry leaves combined. The most effective wavelengths were consistently selected from the red-edge and near-infrared (NIR) spectral regions, which highlights their importance for early detection. To the best of our knowledge, this is the first known study to successfully apply a combined machine learning model for early spider mite detection across two different crop species using hyperspectral data under controlled conditions. The results show the potential of machine learning for multi-crop pest detection and could lay the groundwork for practical, sensor-based tools in precision agriculture.
Predicting herbage biomass on small‐scale farms by combining sward height with different aggregations of weather data
(2024) Scheurer, Luca; Leukel, Joerg; Zimpel, Tobias; Werner, Jessica; Perdana‐Decker, Sari; Dickhoefer, Uta
Accurate predictions of herbage biomass are important for efficient grazing management. Small‐scale farms face challenges using remote sensing technologies due to insufficient resources. This limitation hinders their ability to develop machine learning‐based prediction models. An alternative is to adopt less expensive measurement methods and readily available data such as weather data. This study aimed to examine how different temporal aggregations of weather data combined with compressed sward height (CSH) affect the prediction performance. We considered weather features based on different numbers of weather variables, statistical functions, weather events, and periods. Between 2019 and 2021, data were collected from 11 organic dairy farms in Germany. Herbage biomass exhibited high variability (coefficient of variation [CV] = 0.65). Weather data were obtained from on‐farm and nearby public stations. Prediction models were learned on a training set ( n = 291) and evaluated on a test set ( n = 125). Random forest models performed better than models based on artificial neural networks and support vector regression. Representing weather data by a single feature for leaf wetness reduced the root mean square error (RMSE) by 12.1% (from 536 to 471 kg DM ha −1 , where DM is dry matter) and increased the R 2 by 0.109 (from 0.518 to 0.627). Adding features based on multiple variables, functions, events, and periods resulted in a further reduction in RMSE by 15.9% ( R 2 = 0.737). Overall, different aggregations of weather data enhanced the accuracy of CSH‐based models. These aggregations do not cause additional effort for data collection and, therefore, should be integrated into CSH‐based models for small‐scale farms.
Technical evaluation of a solar-biomass flatbed dryer for maize cobs drying in Rwanda
(2023) Ntwali, Janvier; Romuli, Sebastian; Bonzi, Joévin Wiomou; Müller, Joachim
The persistent problem of postharvest losses in the maize value chain poses an arduous challenge for smallholder farmers in Rwanda, ultimately reducing their market bargaining power. As a consequence, there is an exacerbated disparity in revenues that makes farmers, predominantly female farmers, more vulnerable. The existing drying facilities are based on ambient air drying with a long drying time and the alternative mechanical dryers use mostly fossil fuels which is not a sustainable solution. A solar-biomass hybrid flatbed dryer for maize cobs drying was designed and constructed in the high-altitude volcanic zone of Rwanda. The objective was to provide farmers with an affordable and sustainable drying system with a high drying rate compared to the existing method. In this study, we present the results of the technical evaluation of the dryer to rate its capacity to dry maize cobs to the recommended moisture content. Energy balance was assessed by temperature sensors, airflow distribution was measured with a vane anemometer and the solar radiation from weather station were compared to the solar system data recorded through a datalogging charge controller. Maize was dried in three batches and the moisture content was measure with oven method. Results showed a uniform distribution of airflow on the dryer perforated flow. The burner consumed on average 6 kg of empty cobs per hour and the burner efficiency was 59.4 %. The solar system provided a maximum daily yield of 2.6 kWh, and the battery was able to maintain the system during days of low solar energy availability. Maize cobs were dried from an average moisture content of 23.0 % to 13.7 % in an average period of 90.6 hours. This drying time was significantly lower compared to the already existing system which uses more than 6 weeks. The results prove that the solar-Biomass hybrid flatbed dryer was appropriate for drying maize cobs to the recommended moisture content and thus reduce the risk of postharvest losses in maize value chain in Rwanda. The dryer might be further improved by combining the burner with a solar heating system to further reduce the biomass mass consumption.

Browse

Browsing Institut für Agrartechnik by Sustainable Development Goals "2"