Institut für Phytomedizin

Permanent URI for this collectionhttps://hohpublica.uni-hohenheim.de/handle/123456789/14

Browse

Now showing 1 - 4 of 4

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.
Efficacy of various mechanical weeding methods - single and in combination - in terms of different field conditions and weed densities
(2021) Naruhn, Georg-Peter; Peteinatos, Gerassimos G.; Butz, Andreas F.; Möller, Kurt; Gerhards, Roland
Public awareness and environmental policies have increased interest in applying non-herbicide weed control methods in conventional farming systems. Even though mechanical weed control has been used for centuries in agricultural practice, continuous developments—both in terms of implements and automation technologies—are continuously improving the potential outcomes. Current mechanical weed control methods were evaluated for their weed control efficacy and effects on yield potential against their equivalent herbicide methods. Furthermore, not much is known about the correlation between weed control efficacy (WCE) of different mechanical methods at varying weed density levels. A total of six experiments in winter wheat (2), peas (2), and soybean (2) were carried out in the years 2018, 2019, and 2020 in southwestern Germany. Harrowing and hoeing treatments at different speeds were carried out and compared to the herbicide treatments and untreated control plots. Regarding the average WCE, the combination of harrowing and hoeing was both the strongest (82%) and the most stable (74–100%) mechanical treatment in the different weed density levels. Whereas, in average, hoeing (72%) and harrowing (71%) were on the same WCE level, but harrowing (49–82%) was more stable than hoeing (40–99%). The grain yields in winter wheat varied between 4.1 Mg∙ha−1 (control) and 6.3 Mg∙ha−1 (harrow), in pea between 2.8 Mg∙ha−1 (hoe slow) and 5.7 Mg∙ha−1 (hoe fast) and in soybean between 1.7 Mg∙ha−1 (control) and 4 Mg∙ha−1 (herbicide). However, there were no significant differences in most cases. The results have shown that it is not possible to pinpoint a specific type of treatment as the most appropriate method for this cultivation, across all of the different circumstances. Different field and weather conditions can heavily affect and impact the expected outcome, giving, each time, an advantage for a specific type of treatment.
Enhancing weed suppression in plants by artificial stress induction
(2025) Merkle, Michael; Petschenka, Georg; Belz, Regina; Gerhards, Roland
Various plant species from the Poaceae, Cannabaceae, and Brassicaceae families are used as cover crops to suppress weeds and volunteer crops through competition and allelopathy. This study examined the effects of artificially induced stress on the physiological processes, total phenolic content (TPC), and allelopathic potential of the plant species Avena strigosa, Cannabis sativa , and Sinapis alba at an early growth stage with the aim to increase their weed suppression abilities. Stress was induced at the 3–4 leaf stage in greenhouse-grown plants via harrowing, methyl jasmonate (MeJA) application, insect stress simulation, or a combination of insect stress and harrowing. Maximum quantum yield of photosystem II and shoot dry matter in the three plant species were only minimally or not affected a few days after treatment (DAT). Insect stress caused visible symptoms on treated leaves in all plants. The TPC in the shoot extracts of combined stress-treated C. sativa and insect-stressed S. alba was significantly higher by 1.7 and 1.9 times, respectively, five DAT compared to the shoot extracts from untreated control plants. Additionally, laboratory bioassays with aqueous shoot extracts from the untreated and treated plants were conducted to identify changes in allelopathic potential within the shoot tissues. The application of shoot extracts from MeJA-treated C. sativa and S. alba resulted in the lowest seed germination rates for the two weed species Alopecurus myosuroides and Stellaria media , as well as for the volunteer wheat Triticum aestivum , which were up to 65% lower 10 DAT compared to seeds treated with shoot extracts from non-stressed plants. However, the root-suppressing effect of the shoot extracts on weeds was not influenced by the stress treatments. This study reveals that artificial stress induction can be a suitable management strategy to enhance weed and volunteer cereal suppression in plants in an early growth stage but may vary between stress types and plant species, and requires further optimization and field testing.
Extracellular vesicles isolated from dsRNA-sprayed barley plants exhibit no growth inhibition or gene silencing in Fusarium graminearum
(2022) Schlemmer, Timo; Lischka, Richard; Wegner, Linus; Ehlers, Katrin; Biedenkopf, Dagmar; Koch, Aline
Numerous reports have shown that incorporating a double-stranded RNA (dsRNA)-expressing transgene into plants or applying dsRNA by spraying it onto their leaves successfully protects them against invading pathogens exploiting the mechanism of RNA interference (RNAi). How dsRNAs or siRNAs are transferred between donor host cells and recipient fungal cells is largely unknown. It is speculated that plant extracellular vesicles (EVs) function as RNA shuttles between plants and their pathogens. Recently, we found that EVs isolated from host-induced gene silencing (HIGS) or spray-induced gene silencing (SIGS) plants contained dsRNA-derived siRNAs. In this study, we evaluated whether isolated EVs from dsRNA-sprayed barley ( Hordeum vulgare ) plants affected the growth of the phytopathogenic ascomycete Fusarium graminearum . Encouraged by our previous finding that dropping barley-derived EVs on F. graminearum cultures caused fungal stress phenotypes, we conducted an in vitro growth experiment in microtiter plates where we co-cultivated F. graminearum with plant EVs isolated from dsRNA-sprayed barley leaves. We observed that co-cultivation of F. graminearum macroconidia with barley EVs did not affect fungal growth. Furthermore, plant EVs containing SIGS-derived siRNA appeared not to affect F. graminearum growth and showed no gene silencing activity on F. graminearum CYP51 genes. Based on our findings, we concluded that either the amount of SIGS-derived siRNA was insufficient to induce target gene silencing in F. graminearum, indicating that the role of EVs in SIGS is minor, or that F. graminearum uptake of plant EVs from liquid cultures was inefficient or impossible.

Browse

Browsing Institut für Phytomedizin by Sustainable Development Goals "12"