Agaricus Bisporus Growth Status Prediction in Digital Twin Smart Farm

In recent years, digital twin has gradually begun to be promoted in the field of agriculture, and the establishment of an agricultural digital twin system is able to simultaneously realize the following: digital scene mapping driven by physical monitoring data in real time; remote control of the physical scene by the digital scene; and monitoring and utilization of multisource dynamic data in the life cycle of crops to provide a theoretical basis for decision-making in the planting process. In this paper, based on the function of the digital twin system growth scene projection, due to the intensive growth characteristics of Agaricus bisporus, a spatio-temporal prediction algorithm of Agaricus bisporus growth state image is proposed. Currently, there are fewer image prediction algorithms for the growth state of dense-growing crops such as Agaricus bisporus, and the algorithm proposed in this paper considers the spatial location information contained in the image, integrates the self-attention module, generative adversarial network and spatio-temporal prediction model, and realizes better growth prediction accuracy, to guide the harvesting of Agaricus bisporus in digital twin smart farm.