This dataset provides a comprehensive 10-m resolution vector map of utility-scale and rooftop photovoltaic (PV) power plants across China for the year 2024. It was generated using a novel framework that integrates a Customized Hybrid Genetic Algorithm (CHGA) for feature optimization with a scalable LightGBM classifier, applied to multi-temporal Sentinel-1 SAR and Sentinel-2 optical imagery. The extraction model achieved a high overall accuracy of 95.16%. The dataset is released in ESRI Shapefile format (WGS-84 coordinate system) and includes polygon geometries representing individual PV installations. Each polygon is attributed with estimated installed capacity, annual electricity generation, and avoided CO₂ emissions, derived using consistent provincial grid emission factors. To ensure full transparency and reproducibility, the complete model framework, Sentinel-1/2 preprocessing pipelines, and post-processing code are released alongside the dataset. This resource is intended to support research in spatial energy planning, decarbonization potential assessment, and climate resilience analysis for PV infrastructure.