Phenotyping of genetic resources is an important prerequisite for the selection of resistant varieties in breeding programs and research. Computer vision techniques have proven to be a useful tool for digital phenotyping of diseases of interest. One pathogen that is increasingly observed in Europe is Diplocarpon coronariae, which causes apple blotch disease. In this study, a high-throughput phenotyping method was established to evaluate genetic apple resources for susceptibility to D. coronariae. For this purpose, inoculation trials with D. coronariae were performed in a laboratory and images of infested leaves were taken 7, 9 and 13 days post inoculation. A pre-trained YOLOv5s model was chosen to establish the model, which was trained with an image dataset of 927 RGB images. The images had a size of 768 × 768 pixels and were divided into 738 annotated training images, 78 validation images and 111 background images without symptoms. The accuracy of symptom prediction with the trained model was 95%. These results indicate that our model can accurately and efficiently detect spots with acervuli on detached apple leaves. Object detection can therefore be used for digital phenotyping of detached leaf assays to assess the susceptibility to D. coronariae in a laboratory.

The evaluation of fruit genetic resources regarding a resistance to pathogens is an essential basis for subsequent selection in fruit breeding. Both genetic analysis and phenotyping of defined traits are important tools and provide decision data in the evaluation process. However, the phenotyping of plants is often carried out ‘by hand’ and remains the bottleneck in fruit breeding and fruit growing. The development of a digital and UAV (unmanned aerial vehicle)-based phenotyping method for the assessment of genotype-specific susceptibility or resistance against diseases in orchards would significantly increase the efficiency of plant breeding. In this framework, a workflow for drone-based monitoring of pathogens in orchards was developed using the European pear rust (Gymnosporangium sabinae) as model pathogen. Pear rust is widespread in orchards and causes conspicuous, clearly visible, yellow to orange-colored disease symptoms.

The monitoring of plant diseases in nurseries, breeding farms and orchards is essential for maintaining plant health. Fire blight (Erwinia amylovora) is still one of the most dangerous diseases in fruit production, as it can spread epidemically and cause enormous economic damage. All measures are therefore aimed at preventing the spread of the pathogen in the orchard and containing an infection at an early stage [1-6]. Efficiency in plant disease control benefits from the development of a digital monitoring system if the spatial and temporal resolution of disease monitoring in orchards can be increased [7]. In this context, a digital disease monitoring system for fire blight based on RGB images was developed for orchards. Between 2021 and 2024, data was collected on nine dates under different weather conditions and with different cameras. The data source locations in Germany were the experimental orchard of the Julius Kühn Institute (JKI), Institute of Plant Protection in Fruit Crops and Viticulture in Dossenheim, the experimental greenhouse of the Julius Kühn Institute for Resistance Research and Stress Tolerance in Quedlinburg and the experimental orchard of the JKI for Breeding Research on Fruit Crops located in Dresden-Pillnitz. The RGB images were taken on different apple genotypes after artificial inoculation with Erwinia amylovora, including cultivars, wild species and progeny from breeding. The presented ERWIAM dataset contains manually labelled RGB images with a size of 1280 × 1280 pixels of fire blight infected shoots, flowers and leaves in different stages of development as well as background images without symptoms. In addition, symptoms of other plant diseases were acquired and integrated into the ERWIAM dataset as a separate class. Each fire blight symptom was annotated with the Computer Vision Annotation Tool (CVAT [8]) using 2-point annotations (bounding boxes) and presented in YOLO 1.1 format (.txt files). The dataset contains a total of 1611 annotated images and 87 background images. This dataset can be used as a resource for researchers and developers working on digital systems for plant disease monitoring.

Perennials are usually pre-cultivated on vegetation mats consisting of coconut fiber (coir), which require weather-dependent irrigation and regular fertilization with fast-acting fertilizer to achieve a saleable condition as quickly as possible. In the pre-cultivation of sheep’s wool–coir–vegetation mats, nitrogen (N) is already sufficiently contained in the vegetation mats due to the natural nitrogen content of the sheep’s wool fibers, so that additional liquid fertilization during pre-cultivation can be dispensed with if necessary. In this study, sheep’s wool–coir–vegetation mats of 4.5 kg/m² were pre-cultivated with 16 perennial plants (8 species) in 2018. Variant 1 (V1) received regular fertilization with mineral liquid fertilizer (total 8.7 g N/m²) during pre-cultivation. Variant 2 (V2) was not fertilized during pre-cultivation. In spring 2019, all pre-cultivated vegetation mats were lifted and laid on an area prepared with topsoil. No additional fertilization was applied after laying. The overall impression, plant height, number of flowering perennials, and plant coverage were examined in the 2018 and 2019 growing seasons, with only minor differences observed between V1 and V2. The number of flowers, biomass, and nitrogen content were determined for the two aster species used. There were differences between V1 and V2 in 2018, but not in 2019. The coverage of perennials of 50%, which is the prerequisite for the saleability of the vegetation mats, was already achieved on both V1 and V2 after 4 months of pre-cultivation. The overall impression of the perennials on both V1 and V2 also did not differ during pre-cultivation nor in the following year. Therefore, liquid fertilization is not necessary during the pre-cultivation of perennials on sheep’s wool–coir–vegetation mats.

Computer vision techniques offer promising tools for disease detection in orchards and can enable effective phenotyping for the selection of resistant cultivars in breeding programmes and research. In this study, a digital phenotyping system for disease detection and monitoring was developed using drones, object detection and photogrammetry, focusing on European pear rust (Gymnosporangium sabinae) as a model pathogen. High-resolution RGB images from ten low-altitude drone flights were collected in 2021, 2022 and 2023. A total of 16,251 annotations of leaves with pear rust symptoms were created on 584 images using the Computer Vision Annotation Tool (CVAT). The YOLO algorithm was used for the automatic detection of symptoms. A novel photogrammetric approach using Agisoft’s Metashape Professional software ensured the accurate localisation of symptoms. The geographic information system software QGIS calculated the infestation intensity per tree based on the canopy areas. This drone-based phenotyping system shows promising results and could considerably simplify the tasks involved in fruit breeding research.