Fig 1.
Research framework.
Table 1.
The basic situation of the four campuses in this study.
Fig 2.
Overview of the study area for the present study: (a) location of Wuxi, Nanjing, Hangzhou, and Shanghai, (b) Lihu campus of Jiangnan University in Wuxi, (c) Zijingang campus of Zhejiang University, (d) Xianlin campus of Nanjing University, and (e) Minhang campus of East China Normal University.
Basemap from OpenStreetMap (https://www.openstreetmap.org).
Table 2.
Valid street-view image statistics.
Fig 3.
Example of BSVIs collection process: (a) road network of JNU, (b) generated sampling points interval of 50 meters, and (c) the result of BSVIs collection.
Basemap from OpenStreetMap (https://www.openstreetmap.org); image taken by the author.
Fig 4.
Segmentation performance based on DeepLabV3+ model and calculation processing: (a) loss function curve in training process based on Cityscapes dataset, (b) example of segmentation results in campus walking space, and (c) calculation process of BSVIs.
Image taken by the author.
Table 3.
The question items utilized to measure four perceptual dimensions.
Table 4.
Performance of machine learning models.
Table 5.
Mean scores, standard deviation of each campus walking perceptions.
Fig 5.
Analysis of the four dimensions of walking perception on each campus.
Basemap from OpenStreetMap (https://www.openstreetmap.org).
Table 6.
Top 10 visual features identified following segmentation of the BSVIs.
Table 7.
Results of regression analysis for the visual features and perception scores.