The precise determination of stellar atmospheric parameters (effective temperature T_eff_, surface gravity log g, and metallicity [Fe/H]) serves as a cornerstone for Galactic studies. This work aims to develop a novel deep learning-based approach, the Atmospheric CSWin Framework (ACF), to measure these parameters with high precision. The ACF employs a dual-input architecture that combines astrometric data (parallaxes and their corresponding errors) from Gaia Early Data Release 3 with photometric images from the fourth data release (DR4) of the SkyMapper Southern Survey (SMSS). The framework utilizes a CSWin Transformer backbone for hierarchical feature extraction from photometric images, integrated with Monte Carlo dropout in the prediction module for robust uncertainty quantification. Trained on cross-matched stars between SMSS DR4 and the third data release of the Galactic Archaeology with HERMES spectroscopic survey, ACF achieves parameter estimates with dispersions of 95.02K for T_eff_, 0.07dex for logg, and 0.14dex for [Fe/H]. Systematic experiments demonstrate: (1) Incorporating parallax information significantly improves the precision of all three parameters, especially logg; (2) Our image-based methods outperform traditional approaches relying on stellar magnitudes or colors, with improvements ranging from 2% to 14%; (3) The ACF achieves parameter estimates approaching those of high-resolution spectroscopic analyses; (4) Our framework remains effective even for low-quality samples, showcasing its robustness and generalizability. Using the ACF, we compiled a comprehensive catalog of atmospheric parameters for one million SMSS DR4 stars.

Cone search capability for table J/A+A/698/A322/catalog (Catalog of stellar atmospheric parameters)