Quantifying White Space Zone of Image Segmentation Task Based on Mask2Former

White space is pivotal in forming and embodying the composition and artistic conception of Chinese classical landscape painting, and it also serves as a key element contrasting sharply with Western landscape painting. This study employs Golden Ratio - a universally recognized aesthetic principle - to determine whether Chinese classical landscape painters subconsciously adhered to this rule during their creative process. Therefore, this paper employs a deep learning semantic segmentation model - the Transformer-based Mask2Former - to randomly select 300 renowned Chinese classical landscape paintings as the training set and 261 as the test set. By calculating the pixel ratios of their white space zone and ink splash zone, it finds that 7.28% of the test set paintings approximate Golden Ratio. This finding shows no significant correlation with human-consensus aesthetic proportions. Based on the findings of latent Golden Ratio proportions in Chinese classical landscape paintings, this study concludes that aesthetic similarity does not consistently align with specific, widely accepted aesthetic ratios. However, this algorithmic model provides an effective computational approach for validating research questions specific to Chinese classical landscape painting, offering a deep learning perspective for addressing aesthetic similarity in this genre.