Cross-platform complementarity: Assessing the data quality and availability of Google Street View and Baidu Street View
Top Cited Papers
- 11 February 2025
- journal article
- Published by SAGE Publications in Transactions in Urban Data, Science, and Technology
- Vol. 4 (1) , 22-47
Abstract
Google Street View (GSV) and Baidu Street View (BSV) cover most of the world’s cities. However, their combined use and research application still lack a unified standard. We selected Hong Kong Special Administrative Regions of China (SAR), a region where both types of street view image (SVI) are available, as the study area. In this study, we developed an SVI collection method based on a search-oriented approach. Using an SQLite database to manage over 700,000 SVIs, we conducted comparative experiments. Deep learning methods were employed to process the data and extract visual elements from both types of SVI. We performed a comprehensive comparison of the data quality and usability of the two SVIs in terms of temporal coverage, data acquisition efficiency, acquisition repeatability, and visual element similarity. The findings indicate that BSV outperforms GSV in terms of acquisition efficiency and repeatability, while GSV shows better temporal coverage. The results also confirmed high correlations between GSV and BSV in visual elements related to buildings, greenery, and sky, with correlation coefficients (R) of 0.781, 0.664, and 0.653, respectively. The comparative framework proposed in this study provides theoretical support for geoinformation researchers utilizing SVI data for global-scale studies.Keywords
Funding Information
- the Research Initiative Fund for Newly Introduced Talents of Harbin Institute of Technology, Shenzhen (#ZX20230488)
This publication has 89 references indexed in Scilit:
- Machine Learning Approaches for Measuring Neighborhood Environments in Epidemiologic StudiesCurrent Epidemiology Reports, 2022
- Understanding architecture age and style through deep learningCities, 2022
- Examining the association between the built environment and pedestrian volume using street view imagesCities, 2022
- Human settlement value assessment from a place perspective: Considering human dynamics and perceptions in house price modelingCities, 2021
- Measuring human perceptions of streetscapes to better inform urban renewal: A perspective of scene semantic parsingCities, 2021
- Understanding cities with machine eyes: A review of deep computer vision in urban analyticsCities, 2020
- Semantic Understanding of Scenes Through the ADE20K DatasetInternational Journal of Computer Vision, 2018
- The value of crowdsourced street-level imagery: examining the shifting property regimes of OpenStreetCam and MapillaryGeoJournal, 2018
- Geospatial Data in a Changing World2016
- Using Google Street View to Audit Neighborhood EnvironmentsAmerican Journal of Preventive Medicine, 2011