Understanding the temporal variation of plant community composition is one of the prerequisites to address desertification process and its driving factors and to further develop the effective measures for ecosystem restoration in arid region. Investigating the spatial pattern dynamics of plant communities may help to improve our understandings of vegetation degradation and restoration processes. Till now numerous studies have focused on the changes of community composition during vegetation degradation processes, whereas few studies paid concern to the spatial pattern dynamics of plant communities during the ecological restoration processes. Plants in arid ecosystems are usually sparse with relatively small individual size, which require high-resolution remote sensing data to accurately extract the plant distribution pattern. Recently, the rapid development of unmanned aerial vehicles (UAVs) remote sensing technology has allowed an opportunity for the studies on vegetation spatial pattern variation at fine-scale. In this study, we investigated the spatial pattern changes of plant communities in the ecological restoration sites in the Shapotou area, Ningxia Hui Autonomous Region using both high-resolution remote sensing data of UAVs with 2 cm spatial resolution and ground community survey. The results showed that the implementation of the straw checkerboards in Shapotou has significantly improved the plant species diversity and vegetation coverage compared to those of un-restored areas. After four years of implementation, the average vegetation coverage increased by three times and the species richness increased by one time. During the vegetation restoration process, the vegetation patches have shown a significant transformation in pattern features such as increased patch size, decreased fragmentation rate, complicated shape, and weakened spatial autocorrelation. The changes in these spatial pattern characteristics indicate that (1) The large-scale vegetation patches were in a recovery process. (2) The improvement of microenvironment was conducive to the survival of individual plants. (3) The risk of ecosystem degradation to bare land has been reduced. In this study, we assessed the vegetation spatial pattern variation during the restoration processes in detail using UAV remote sensing technology. Combined with thoroughly ground community survey, the effectiveness of straw checkerboards has been proven in multiple perspectives. The continuous long-term monitoring of the system pattern based on UAVs is valuable and necessary for understanding the ecological restoration mechanism in arid region and for future science-based ecological restoration measures in desertification control.