Fuente: WIPO "processing tomato" The invention provides a greenhouse tomato root length density prediction method and device and a storage medium, and relates to the technical field of data processing for the prediction purpose, and the method comprises the steps: obtaining root length density influence factor data; the root length density influence factor data are input into a pre-trained root length density prediction model, a root length density prediction value output by the root length density prediction model is obtained, and the root length density prediction model is obtained based on the root length density prediction value and attention weight training of each root length density influence factor. According to the greenhouse tomato root length density prediction method and device and the storage medium provided by the invention, the root length density influence factor data is input into the pre-trained root length density prediction model, and the root length density prediction value can be obtained through the semi-supervised generative adversarial network and the depth separable graph convolutional network; the prediction precision of the greenhouse tomato root length density can be improved.