In response to the low level of informatization and time-consuming manual measurement of mushroom phenotype, this paper proposes a U-Net mushroom phenotype extraction method. This method collects cross-sectional images of shiitake mushrooms, establishes a dataset, and achieves segmentation of shiitake mushroom caps, stems, and left and right gills based on U-Net. The average intersection to union ratio and average pixel accuracy of the model are 85.00% and 91.25%, respectively. The automatic measurement of five phenotypic parameters of shiitake mushrooms, including cap diameter, cap thickness, stem length, stem diameter, and gill width, was achieved by combining the centroid method and the minimum bounding rectangle method. Compared with manual measurements, the method proposed in this article has average absolute percentage errors of 1.57%, 5.01%, 2.57%, 5.47%, and 2.74% in measuring cap diameter, cap thickness, stem length, stem diameter, and gill width, respectively; The root mean square errors are 0.12cm,