In this paper, deep learning is mainly introduced into remote sensing image processing, and the convolutional neural network model is used to process remote sensing images, so that the remote sensing images can be classified after the crops are harvested and after harvesting, and then the unknown harvested remote sensing images are processed. classification, which is of great significance in the field of agriculture. We can roughly estimate the harvesting and unharvesting of sugarcane through remote sensing images. In this paper, the harvesting situation of sugarcane and remote sensing images are used to train the harvesting situation, and then the harvesting situation of sugarcane is predicted, so as to monitor the harvesting situation of sugarcane. Make the relevant state departments have an advance plan for the import and export of sugar. The second objective of Sustainable Development (SDGs) is to eradicate hunger, food security, improve nutrition and promote sustainable agriculture, and this paper is an experiment in response to the needs of this objective.