Abstract: Objectives: Integrated remote sensing technology primarily captures deformation indicators on the surface of the slope but fails to delineate the internal deformation characteristics clearly. Consequently, it can only facilitate preliminary identification without accurately determining the current stage of the slope or assessing the risk of sliding. To address these limitations, experimental research was conducted. Methods: To investigate these issues, a centrifugal model test of a large-scale accumulation landslide was performed. Deformation characteristic data for both the surface and interior of the slope were obtained through multi-angle imaging, image processing, and observational deduction. This approach established a correlation between the internal deformation and the macroscopic deformation characteristics of the accumulation landslide. The resulting data provide a semiquantitative reference framework for the early identification and warning of landslides using technologies such as aperture radar, lidar, and optical remote sensing to ascertain the deformation stage of landslides. Results: The results show that: 1) During the development of the accumulation landslide, cracks at the trailing edge extend significantly in the horizontal direction, while cracks on the back wall exhibit an arc-shaped tensile development trend, approximately parallel to the slope's alignment. Slip failure in homogeneous soil slopes typically initiates at the slope crest, with overall failure occurring as the sliding surface gradually advances from back to front. 2) A novel method for classifying landslide stages is proposed, which combines the displacementacceleration curve of the accumulation landslide with the soil's over-consolidation ratio. A sudden change in the over-consolidation ratio serves as a segmentation point, with the soil consolidation settlement stage and stress redistribution stage preceding the change, and the slope failure stage following it. Conclusions: By comparing and analyzing typical cases, the results indicate that model test outcomes can aid remote sensing technology in distinguishing the deformation stages of landslides for early identification. This suggests that utilizing the correspondence mapping between the internal deformation of the accumulation landslide and the surface deformation characteristics offers valuable guidance for interpreting the deformation stage of the landslide through remote sensing.