Plausible or misleading? Evaluating the adaption of the place pulse 2.0 dataset for predicting subjective perception in Chinese urban landscapes

https://doi.org/10.1016/j.habitatint.2025.103333

Abstract

Visual perception is crucial in human-centric spatial studies. Currently, Place Pulse dataset is widely used for subjective scoring of urban space. However, its local applicability raises questions due to limitations in data sources and participants. This study compares the performance of Place Pulse 2.0 and a local dataset from Shenzhen in predicting perceptions, exploring its feasibility for evaluating Chinese megacities. Street view images (SVIs) in Shenzhen were categorized into five spatial types using k-means clustering for refined differentiation, and 400 SVIs from each category were compiled into a dataset rated by local residents. Perception scores based on different pre-trained datasets were predicted using XGBoost. The study found notable differences in urban perceptual evaluation between the datasets. Place Pulse dataset tended to give more pessimistic ratings for negative perceptions like “boring” and “depressing”. For positive perceptions such as lively and safe, it performed moderately, but well in beautiful and wealthy. Additionally, perceptual differences were most pronounced in urban core, green corridors, and urban villages. Urban villages saw an increase in perceptions of being "depressing" due to the homogeneity of building facades and a sense of enclosure. The novelty lies in quantifying perceptual differences across spatial clusters and elaborating their relationship with the visual environment. This research challenges the universality of the Place Pulse dataset in global cities and provides a more credible database and framework for localized spatial perception research in China.

📍 Background

This research evaluates the Place Pulse 2.0 dataset, a widely used tool for understanding subjective perceptions of urban landscapes, and its suitability for predicting these perceptions in the context of Chinese megacities, specifically Shenzhen. While Place Pulse provides valuable data, it suffers from limitations such as cultural and geographical biases due to its origins in Western countries and lack of representation from cities like those in Mainland China. This study critically examines the applicability of Place Pulse 2.0 for urban perception analysis in China.

🎯 Research Purpose

The primary aim of this study is to assess the reliability of Place Pulse 2.0 when applied to the Chinese context, specifically in Shenzhen, and compare its predictions against local datasets generated through subjective surveys. By examining discrepancies across six perceptual indices—safety, liveliness, beauty, wealth, boredom, and depression—the study challenges the generalizability of Place Pulse and advocates for more localized datasets in urban perception research.

🧠 Methodology

1. Data Collection: Street View Images (SVIs) were collected from Shenzhen using Baidu’s API, with a focus on various urban spaces categorized into five distinct clusters through k-means clustering. Local residents provided subjective perception scores on these images via an online survey.
2. Comparative Analysis: The study employs XGBoost models to predict perception scores using both Place Pulse and the local Shenzhen dataset, followed by a comparison of the perceptual distributions and spatial patterns across different urban regions.
3. Semantic Segmentation: DeepLabV3+ was used for image segmentation to extract objective features from the street views, such as greenness, walkability, enclosure, and more, to better understand how these factors influence perceptions.

📊 Key Findings

1. Discrepancies in Perception Scores:
• Place Pulse 2.0 tended to give more pessimistic ratings for negative perceptions like "boring" and "depressing" compared to local perceptions, especially in areas such as urban villages and green corridors.
• For positive perceptions such as lively and safe, Place Pulse performed moderately well but still showed notable differences compared to local residents' evaluations.
• Urban villages and expressways exhibited the greatest perceptual discrepancies, where Place Pulse underestimated the vibrant and safe nature of urban villages.
2. Geographical Variations: The study found significant differences in how perceptions were distributed across spatial clusters. Urban core areas scored higher on wealth and beauty indices, while green corridors were associated with a strong sense of safety and liveliness.
3. Model Accuracy: The XGBoost model provided high accuracy in predicting subjective perception scores, with an R² value of 89.6% for the local dataset, suggesting that localized data can enhance the accuracy of urban perception models.

🌏 Implications

1. Global vs. Local Perception Models: The study underscores the importance of local data in urban perception analysis, especially when dealing with cities that have unique sociocultural and architectural characteristics. Place Pulse’s reliance on non-Asian, Western volunteer data limits its applicability in Chinese cities, particularly in densely populated and culturally diverse areas like Shenzhen.
2. Urban Planning: This research provides valuable insights for urban planners and designers in Shenzhen and similar cities. By understanding the perceptual differences, planners can better cater to local preferences, improving walkability, safety, and green spaces, and avoiding planning biases from using global datasets.
3. Future Directions: The study promotes further development of localized perception datasets, advocating for broader regional participation to mitigate biases in global datasets like Place Pulse.

🏁 Conclusion

This project demonstrates that while Place Pulse 2.0 provides a robust foundation for global urban perception studies, it is misleading when applied to the specific context of Chinese megacities like Shenzhen. By comparing Place Pulse with a localized Shenzhen dataset, the study reveals significant perceptual biases and highlights the importance of incorporating local cultural and spatial factors in urban perception research.