A Large-Scale Multipurpose Benchmark Dataset and Real-Time Interpretation Platform Based on Chinese Rural Buildings

Weihuan Deng,Weipan Xu,Yaofu Huang,Xun Li

DOI: https://doi.org/10.1109/jstars.2024.3386830

IF: 4.715

2024-01-01

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing

Abstract:As urbanization accelerates, the evolving dynamics of village growth and decline have garnered widespread attention. Rural housing, as the most significant asset in villages, serves as the primary indicator of socioeconomic development in rural areas. However, the extensive scale, diversity, and widespread distribution of villages make conducting a nationwide census of rural buildings a notably costly and time-intensive endeavor. Although deep-learning techniques have been successfully applied by numerous researchers to map building footprints, the majority of this work is concentrated in urban areas, leaving large-scale datasets for rural buildings notably lacking. In this article, an exhaustive database of rural architecture has been established, featuring diverse rural building annotations from the majority of provinces in the mainland China. Moreover, a real-time online platform for remote sensing image interpretation, integrating instance segmentation and boundary regularization, has been developed to streamline the extraction of building footprints from high-resolution imagery. The experimental results from predicting 43 992 rural building instances nationwide demonstrated that 33 210 were accurately identified, achieving a precision of 0.776, a recall of 0.755, and an F1-score of 0.765. Building upon this work, the maps of rural building areas and quantity are produced to clearly demonstrate the distribution of rural houses in parts of China. These data products can serve as vital supplements to public data products, such as nighttime light data, land cover maps, national statistical yearbooks, and road network data, particularly in the field of rural studies.