Institut für Züchtungsforschung an gartenbaulichen Kulturen
Every scientist, even those adhering to the highest standards of scientific ethics, is subject to bias during data collection in their experiments, which tends to skew results towards the optimal outcomes of the working hypothesis. Therefore, the goal of this project is to automate the data collection and scoring of phenotypic traits to eliminate human influence and bias. In the first step of the project, an image pool of the desired crop, in this case, Solanum lycopersicum (tomato), will be created. This image pool should contain a sufficient number of labeled images of the traits. Subsequently, a model will be trained and validated using the collected image data. In the final step, this model will be implemented in hardware (e.g., camera, drone, handheld device). By continuously augmenting the image database, it will be possible to recognize and identify not only various phenotypic traits but also deficiency and disease symptoms.
Federal Ministry of Food and Agriculture
