Understanding the Visual Relationship between Function and Facade in Historic Buildings Using Deep Learning—A Case Study of the Chinese Eastern Railway

Abstract

Although functional identifiability represents a key aspect for promoting visual connotation and sustainable usability in historic building groups, there is still no consensus on how to quantitatively describe its identification basis at a large scale. The recent emergence of the potentiality of deep learning and computer vision has provided an alternative to traditional empirical-based judgment, which is limited by its subjective bias and high traversal costs. To address these challenges, this study aims to build a workflow for a visual analysis of function and facade to extract the different contributions that facade elements provide to functional expression. The approach is demonstrated with an experiment on a section of the Chinese Eastern Railway (CER) where large-scale historical buildings images were categorized to identify functions using deep learning, together with activation and substance for visual calculations. First, the dataset aggregated with images of historic buildings along the CER was used to identify functional categories using SE-DenseNet merging channel attention. The results of the model visualized using t-SNE and Grad-CAM were then used to analyze the relationships of facade features across functional categories and differences in elemental feature representation across functional prototypes. The results show the following: (1) SE-Densenet can more efficiently identify building functions from the closely linked facade images of historic building groups, with the average accuracy reaching 85.84%. (2) Urban–rural differences exist not only in the count of spatial distributions among the CER’s historic building groups, but also in a significant visual divergence between functions related to urban life and those involved in the military, industry, and railways. (3) Windows and walls occupy areas with more characteristics, but their decorative elements have a higher intensity of features. The findings could enhance the objective understanding and deeper characteristics of the historical building group system, contributing to integrated conservation and characteristic sustainability.

Mapping Green View Index for Urban Parks with Varied Landscape Metrics and Distances toward the Chinese Eastern Railway Network

“…However, the GVI has been employed in most of the current studies to assess a single city [32,33], and less is known about the GVI with regard to regional areas of special interest. Areas covered by the CER network have been mapped to analyze the spatial distributions of different types of urban infrastructures [3,5]. This supports the idea that regional distributions of landscape surface functions can be characterized by patterns framed with distances to railway lines and landscape metrics.…”

“…A recent study revealed that industrial parks are a type of urban infrastructure that attracts urban planning with vegetation, maintaining a GVI of over 20%. The functions of old infrastructure in cities near CER lines were mainly intended for civil uses, such as business, medicine, education, and religion, among others [5]. There has been little evidence demonstrating the establishment of industrial parks in regions near CER routes.…”

“…The conservation of railway heritage is challenged by abandonment and urbanization [5]. It is contradictory that relic infrastructures may have lost their functions in previous eras; yet, they cannot be directly rebuilt because many of them are regarded as architectural masterpieces [6].…”

“…According to , the CER can be characterized as architectural heritage, encompassing 92 stations covering five levels of regions across the whole area [3]. Recently, people have begun characterizing architectural heritage by recognizing building facades and mapping their distributions [3,5]. Relatively few studies have documented the landscape characteristics of ecological infrastructures along CER routes.…”