Digital Clones at the Adaptable Control in the Agricultural Biotechnology

In the article, results of the development of methods, models, and hardware/software solutions for creating and actualization of digital clones of crops with the complex structure in the form of a complex of 3D models are presented that ensure the possibility to perform virtual biological experiments consisting in the cultivation of agricultural plants in the context of in vitro conditions (in a test glass) with evaluation and forecasting of parameters that have an effect for the further field setting and adaptation of plants in conditions of the outdoor bed, and prevailing natural environment and climatic factors. For the segmentation of the plant using methods of machine learning, a segmenting neuron net with the U2 –Net architecture was used. Good results of learning were obtained. A prototype of an automated installation has been developed that makes it possible to perform the complete cycle of the digital phenotyping and the analysis of obtained results based on digital clones of plants and implementation of the virtual process of in vitro cultivation. The obtained complex makes it possible to perform studies in that the microclimate inside of the test glass will not be disrupted; the data registration process is accelerated essentially; the human factor and the subjectivity are excluded during measurements. The knowledge base has been created that includes 792 units of 3D models for six crop species.