Die chronologische Liste zeigt aktuelle Veröffentlichungen aus dem Forschungsbetrieb der Hochschule Weihenstephan-Triesdorf. Zuständig ist das Zentrum für Forschung und Wissenstransfer (ZFW).
8 Ergebnisse
M.Sc. Benedikt Thesing,
Peter Weindl,
M.Sc. Sina Göppel,
Sebastian Born,
P. Hofmann,
Christian Lambertz,
Prof. Dr. Gerhard Bellof
Weeds are undesired plants in agricultural fields that affect crop yield and quality by competing for nutrients, water, sunlight and space. For centuries, farmers have used several strategies and resources to remove weeds. The use of herbicide is still the most common control strategy. To reduce the amount of herbicide and impact caused by uniform spraying, site-specific weed management (SSWM) through variable rate herbicide application and mechanical weed control have long been recommended. To implement such precise strategies, accurate detection and classification of weeds in crop fields is a crucial first step. Due to the phenotypic similarity between some weeds and crops as well as changing weather conditions, it is challenging to design an automated system for general weed detection. For efficiency, unmanned aerial vehicles (UAV) are commonly used for image capturing. However, high wind pressure and different drone settings have a severe effect on the capturing quality, what potentially results in degraded images, e.g., due to motion blur. In this paper, we investigate the generalization capabilities of Deep Learning methods for early weed detection in sorghum fields under such challenging capturing conditions. For this purpose, we developed weed segmentation models using three different state-of-the-art Deep Learning architectures in combination with residual neural networks as feature extractors.We further publish a manually annotated and expert-curated UAV imagery dataset for weed detection in sorghum fields under challenging conditions. Our results show that our trained models generalize well regarding the detection of weeds, even for degraded captures due to motion blur. An UNet-like architecture with a ResNet-34 feature extractor achieved an F1-score of over 89 % on a hold-out test-set. Further analysis indicate that the trained model performed well in predicting the general plant shape, while most misclassifications appeared at borders of the plants. Beyond that, our approach can detect intra-row weeds without additional information as well as partly occluded plants in contrast to existing research.All data, including the newly generated and annotated UAV imagery dataset, and code is publicly available on GitHub: https://github.com/grimmlab/UAVWeedSegmentation and Mendeley Data: https://doi.org/10.17632/4hh45vkp38.3
Mehr
Maura John,
Prof. Dr. Florian Haselbeck,
Rupashree Dass,
Christoph Malisi,
Christian Dreischer,
Sebastian J Schultheiss,
Prof. Dr. Dominik Grimm
Genomic selection is an integral tool for breeders to accurately select plants directly from genotype data leading to faster and more resource-efficient breeding programs. Several prediction methods have been established in the last few years. These range from classical linear mixed models to complex non-linear machine learning approaches, such as Support Vector Regression, and modern deep learning-based architectures. Many of these methods have been extensively evaluated on different crop species with varying outcomes. In this work, our aim is to systematically compare twelve different phenotype prediction models, including basic genomic selection methods to more advanced deep learning-based techniques. More importantly, we assess the performance of these models on simulated phenotype data as well as on real-world data from Arabidopsis thaliana and two breeding datasets from soy and corn. The synthetic phenotypic data allows us to analyze all prediction models and especially the selected markers under controlled and predefined settings. We show that Bayes B and linear regression models with sparsity constraints perform best under different simulation settings with respect to explained variance. Further, we can confirm results from other studies that there is no superiority of more complex neural network-based architectures for phenotype prediction compared to well established methods. However, on real-world data, for which several prediction models yield comparable results with slight advantages for Elastic Net, this picture is less clear, suggesting that there is a lot of room for future research.
Gustavo Chaves,
Artem G. Ayuyan,
Vladimir V. Cherny,
Deri Morgan,
Arne Franzen,
Lynne Fieber,
Prof. Dr. Lydia Nausch,
Christian Derst,
Iryna Mahorivska,
Christophe Jardin,
Thomas E. DeCoursey,
Boris Musset
ObjectivesSomatic cell count (SCC) is a reliable and approved parameter for the estimation of udder health. The maincell types regarding somatic cells in the udder are lymphocytes, macrophages, and polymorph nuclearleucocytes (PMN). The differential somatic cell count (DSCC) represents the proportion of lymphocytes andPMN to total SCC, the remaining percentages to SCC are macrophages. So far, the effects of milk yield,season, parity, milking frequency, days in milk, and major pathogens on the DSCC are already described. Afurther known effect on udder health and SCC is the milking interval (MI). On farms with automatic milkingsystems (AMS) the MI can vary for each cow compared to conventional milking systems. Regarding DSCCand SCC, cows milked by AMS systems showed higher values compared to cows milked by conventionalmilking systems. Therefore, the aim of this study was to evaluate the effect of MI on DSCC.Materials and methodsData from 27 dairy herds from Germany including 6,500 dairy cows and 43,229 recordings were evaluated.The data resulting from milk yield performance testing were collected between January to December 2020. Allherds used automatic milking systems combined with free cow traffic. Milk yield performance testing data wasrecorded 11-times per year on each farm and included the DSCC measured using the FOSS methoddescribed by Damm et al. (2017). Date and time from each milking at the farms were available and used tocalculate each individual MI between milkings. MI ranged from 1 h minimum to 24 h maximum. Data of milkinginterval >24 h were excluded. Means were compared using Wilcoxon test. P-values were Bonferroni adjusted;the threshold for significance was set after adjusting to α < 0.05. A linear mixed model was used to estimatethe effect on DSCC including MI, milk yield, lactation, days in milk, and season as fixed effects and herd,individual cow, and residuals as random effects.ResultsMean MI was 10.6 h (±0.04 h SE). MI of ≤4 h resulted in the highest DSCC (52.3 ±1.0%). The DSCCdecreased significantly for cows showing a MI >4 and ≤6 h (39.0 ±0.6%) and had its minimum between MI >6and ≤8 h (37.9 ±0.4%). MI between >8 - ≤10 h resulted in a DSCC of 40.5% (±0.4%). The DSCC increased forMI >10 - ≤12 h and for >12 h MI (42.8 ±0.4%; 46.6 ±0.3%, respectively; all P-values < 0.001). Therefore, themost frequently milked cows showed higher DSCC compared to cows between 4 and 8 hours MI. Consideringnatural behavior, the suckling interval of calves from their mothers ranges between 4 to 6 times per day, whichresults in a MI of 6 to 4 hours, representing the MI of the second lowest DSCC found in this dataset.The standard deviation of the MI (MISD) expresses the irregularity of milkings. Data evaluation showed the lower the MISD, the lower the DSCC. For MISD ≤2 h the DSCC was lowest (38.8 ±0.7%), compared to MISD >2- ≤4 h (41.0 ±0.5%), MISD >4 - ≤6 h (43.2 ±0.7%), and MISD >6 h (48.1 ±1.1%). Irregular milking is also knownto impair udder health and increase the SCC of cows.ConclusionsMilking interval between 4 to 8 hours minimizes DSCC, which aims the natural MI of suckling calves. A moreregular milking interval in AMS farms could reduce DSCC and therefore improve udder health. AMS farmsshould strive their management and settings of the AMS to encourage cows to visit the AMS more regularly.AcknowledgementWe kindly acknowledge the QNETICS GmbH, Erfurt, Germany, for providing the dataset of DSCC values andmilk yield recording data for this study.
ObjectivesTail injuries and pathological alterations have been reported in many species. In cattle, they were investigatedmainly in fattening bulls and feedlot cattle. In dairy cows high prevalences for different tail alterations werefound. However, aetiology and pathogenesis of this health trait are still unclear and need further investigation.Out of 4443 phenotypes of different tail alterations we assorted seven groups common in dairy cows: 1. verytip of the tail , 2. ring-like, 3. scurf, 4. swelling, 5. thinning, 6. axis anomaly, and 7. verruca-like mass. Theobjective of this study was to identify genomic regions that may influence the occurrence of different tailalterations in dairy cows, which could be useful for a potential implementation of a genomic selection tool formore robust and healthy cows in the future.Material and methodsData collection started in December 2019 from a German 75 German Holstein (HOL) cows dairy herd. All cows wereexamined every two weeks during six months regarding any kind of tail alterations. The findings were described andphotographed. Data analysis resulted in seven different kinds of tail alterations: 1. very tip of the tail, 2. ring-like, 3. scurf,4. swelling, 5. thinning, 6. axis anomalies, and 7. verruca-like mass.Hereinafter, prevalences for the observed tail alterations were calculated based on monthly data collection from fivedifferent dairy herds: 3 HOL herds, counting average herd sizes of 75, 300, and 1300, respectively; 2 German Fleckvieh(FV) herds, counting 60 cows, each. All cows were housed in free stall barns with conventional (HOL, FV) or automaticmilking systems (FV).In total, 4443 Dairy Cows' Tail Scores were recorded. Data preparation and analysis were performed using R version4.1.2. Prevalences for tail alterations were calculated by dividing the number of observations within by the total number ofobservations of each kind of tail alteration and was given in percent. For calculating the total prevalence per breed andfarm, the occurrence of at least one tail alteration counted as an observation, was divided by the total number of cowsunder investigation and given in percent.ResultsThe overall prevalence for any kind of tail alteration was 88% in German Holstein and 99% in Fleckvieh cows; it variedbetween farms from 74% to 99%. Prevalences for HOL and FV regarding alterations of the very tip of the tail were 26%and 71%, ring-like alterations 24% and 30%, swelling 26% and 42%, scurf 55% and 60%, thinning combined with axisanomalies 16% and 21%, and verruca-like mass 10% and 21%, respectively. Number per tail ranged for ring-likealterations and thinning/axis anomalies from 1 to 5 and for verruca-like mass from 1 to 3.ConclusionsDuring this study, high prevalences for different tail alterations in HOL and FV dairy cows were found out. The grouping ofdifferent alterations as described above can be useful to phenotype tail alterations in dairy cows. However, furtherinvestigations regarding pathogenesis, aetiology, and genetics of the observed alterations in dairy cows' tails are neededto understand their origin and impact on animal health and welfare.FundingThis research was funded by the Tönnies Forschung, Rheda, Germany.
Mehr
Betreuung der Publikationsseiten
Gerhard Radlmayr
Referent für Wissenstransfer und Forschungskommunikation
Wir verwenden Cookies. Einige sind notwendig für die Funktion der Webseite, andere helfen uns, die Webseite zu verbessern. Um unseren eigenen Ansprüchen beim Datenschutz gerecht zu werden, erfassen wir lediglich anonymisierte Nutzerdaten mit „Matomo“. Um unser Internetangebot für Sie ansprechender zu gestalten, binden wir außerdem externe Inhalte unserer Social-Media-Kanäle ein.