Institut für Züchtungsforschung an Obst
The assessment of fruit load in apple trees plays a crucial role in both accurate yield estimation and the evaluation of biennial bearing. Biennial bearing refers to the year-to-year fluctuation between high and low yields, a well-known issue in apple cultivation that negatively affects both yield stability and fruit quality. Evaluation of biennial bearing in genetic resources allows the identification of apple varieties with a lower tendency towards biennial bearing. These insights are the basis for breeding of new varieties that are stable in yield while possessing both high productivity and improved adaptability.Currently, fruit load assessment is done manually through visual inspections, which is very time-consuming and prone to errors. Therefore, a digital phenotyping method is being developed that enables efficient and highly accurate assessment of fruit load. The goal is to develop an automated, high-throughput method for fruit load evaluation to assess apple genetic resources and in breeding programs.The digital phenotyping method aims to ensure high accuracy in fruit detection, considering both traditional and newly developed apple varieties with diverse fruit shapes and sizes. This technology is expected to significantly improve efficiency and precision in evaluating genetic resources.For this purpose, drone flights will be conducted over the apple variety collection at the Julius Kühn Institute (JKI-ZO) during fruit ripening over at least two years. These flights will capture high-resolution images of the fruit-bearing trees, which will be used to create a image database. The collected image data will be manually annotated by labeling and classifying the fruits in the images using a software program. These annotated data will serve as a training dataset for developing a machine learning model capable of detecting fruit load accurately. Once a satisfactory detection model has been developed, it will be validated on a separate test dataset. This validation process is crucial to ensure the robustness and accuracy of the model, especially given the diversity of fruit sizes, shapes, and colors present in both traditional and newly bred apple varieties.The development of this digital phenotyping method is expected to represent a significant advancement in the assessment of fruit load and biennial bearing in apple genetic resources. Through automated and precise measurement, the efficiency of breeding research will be greatly enhanced, allowing larger datasets to be analyzed in less time. In the long term, this method aims to contribute to the breeding of apple varieties with stable yields, which are less susceptible to biennial bearing while maintaining high fruit quality.
Federal Ministry of Food and Agriculture
