Abstract: Taking Chenggong District of Kunming City as an example, multi-temporal remote sensing images from 2011, 2016, and 2021 obtained from Tianditu were utilized to automatically extract urban construction land using ArcGIS Pro 3.3 and deep learning technologies, resulting in vector datasets for these three periods. Methods such as expansion rate, expansion intensity, compactness, centroid shift, standard deviation ellipse, hotspot analysis, and buffer zone analysis were employed to quantitatively assess the scale, spatial distribution, and changing trends of the newly added construction land. The results indicate that Chenggong District experienced rapid outward expansion from 2011 to 2016, with newly developed areas primarily concentrated in peripheral regions. However, from 2016 to 2021, the expansion rate significantly slowed down, with development focus gradually shifting towards the central and mid-ring areas, reflecting a transformation from extensive expansion to intensive optimization.