Abstract: Guizhou province is confronted with a relative deficiency of arable land resources, limited scope for expanding its cultivated areas,posing a challenge to its regional agricultural production and food security. Timely monitoring of land occupation is of great significance for farmland protection and loss reduction. Remote sensing technology can play an important role in monitoring cropland occupation, however, due to the complexity of the surface structure, high-precision monitoring of cropland occupation faces greater difficulties. In order to improve the monitoring accuracy, this paper investigates the use of deep learning technology to monitor the cultivated land occupation in Guizhou province. Firstly,multi-type and high-frequency high-resolution satellite images are utilized to obtain a large number of samples in the whole area of Guizhou province, according to which the information of cultivated land occupation in remote sensing images is mined. Then,convolutional neural network and recurrent neural network are jointly used to construct a deep learning network model for monitoring cropland changes, and cropland changes are extracted from the spectral, spatial and temporal phase information of remote sensing images. Finally,typical areas are selected to verify the accuracy of the monitoring results. The results show that the method can quickly monitor the areas of occupied cropland in Guizhou province, and provide decision-making references and regulatory tools for relevant departments.

Key words: misappropriation of arable land, deep learning, change detection, remote sensing image, Guizhou province