<p>Street spatial form has long been a concern for urban researchers and planners. Fine-grained measurement of street spatial form provides critical insights for understanding urban environments, thereby supporting spatial analysis and policymaking. However, existing studies struggle to capture three-dimensional street characteristics across large geographic extents. Reliance on traditional surveys or localized simulations often limits analytical scope and result robustness. This study introduces an efficient and scalable measurement framework. It utilizes street view images as a data source and integrates deep learning algorithms with camera projection modeling to automatically extract street dimensions and morphological indicators. We validate the framework through a series of experiments. Taking the Guangdong-Hong Kong-Macao Greater Bay Area as a case study, we present the first regional-scale thematic map of street spatial form, identify representative morphological types, and examine the associations between street spatial form and multidimensional urban environments. This work proposes an innovative perspective for addressing the scarcity of three-dimensional street form data, offers a practical data-driven approach for large-scale streetscape analysis, and highlights the application potential of street form features in urban studies.</p>