<p>High-precision and wide coverage data on rural household wealth are essential for bridging national-level rural revitalization policies with micro-level rural entities, enabling the precise allocation of public resources. However, the vast number and dispersed distribution of rural communities in China make wealth data difficult to collect and scarce in availability. To address this challenge, this study proposes an integrated technical framework that combines “sky” remote sensing imagery with “ground” village street view imagery to construct a fine-grained, computable representation of rural household wealth. Centered on the intelligent interpretation of rural housing features, we extract wealth-related visual elements from imagery and regress them against benchmark survey-based household wealth indicators to develop a high-accuracy township-level wealth prediction model (<i>R</i>² = 71%). This model is used to generate a nationwide, township-scale rural household wealth map. Our findings reveal a distinct “bimodal” spatial distribution of rural wealth in China, characterized by a polarization pattern: higher in the south and east, and lower in the north and west. This approach offers a scalable and cost-effective alternative to traditional household surveys, supporting the identification of rural development gaps and facilitating more targeted and effective rural policy implementation.</p>