Scholarlink

<p>With the accelerated pace of global urbanization, residents’ subjective perception of urban spaces has increasingly become a central issue in urban planning. Traditional research methods typically rely on static street view images, neglecting the visual attention patterns reflected by eye-tracking behavior and their interaction with urban space features in shaping residents’ perceptions. In response, this study proposes a multimodal urban perception model that integrates street view images and eye-tracking data, quantifying residents’ perceptions across six dimensions: Beautiful, Boring, Depressing, Lively, Safe, and Wealthy. Validation on the MIT Place Pulse 2.0 dataset demonstrates that the proposed model significantly outperforms existing methods. Furthermore, using Beijing’s Sixth Ring Road as a case study, a localized dataset of street view images and eye-tracking data is constructed, revealing the relationship between visual attention patterns and spatial perception. The results show that the model integrating eye-tracking data reduces the Mean Absolute Error (MAE) by 19.9%–21.4%, the Mean Squared Error (MSE) by 15.4%–21.8%, and increases the R² coefficient by 35.4%–44.7%, compared to the model that only uses street view images. The urban perception map drawn based on the prediction results of this model reveals the spatial differentiation characteristics of various areas within the Sixth Ring Road of Beijing in different perception dimensions, providing strong data support and a new spatial planning perspective for optimizing urban spatial layout and improving the quality of life of residents.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Urban perception analysis: integrating street view images with eye-tracking data

  • 期刊论文

  • 发表日期：

    2026年2月9日

查看全文
  • Jing Jin,
  • Huifu Li,
  • Rui Zhou,
  • Su Li,
  • Abdureyim Raxidin,
  • Xun Zhang

摘要

With the accelerated pace of global urbanization, residents’ subjective perception of urban spaces has increasingly become a central issue in urban planning. Traditional research methods typically rely on static street view images, neglecting the visual attention patterns reflected by eye-tracking behavior and their interaction with urban space features in shaping residents’ perceptions. In response, this study proposes a multimodal urban perception model that integrates street view images and eye-tracking data, quantifying residents’ perceptions across six dimensions: Beautiful, Boring, Depressing, Lively, Safe, and Wealthy. Validation on the MIT Place Pulse 2.0 dataset demonstrates that the proposed model significantly outperforms existing methods. Furthermore, using Beijing’s Sixth Ring Road as a case study, a localized dataset of street view images and eye-tracking data is constructed, revealing the relationship between visual attention patterns and spatial perception. The results show that the model integrating eye-tracking data reduces the Mean Absolute Error (MAE) by 19.9%–21.4%, the Mean Squared Error (MSE) by 15.4%–21.8%, and increases the R² coefficient by 35.4%–44.7%, compared to the model that only uses street view images. The urban perception map drawn based on the prediction results of this model reveals the spatial differentiation characteristics of various areas within the Sixth Ring Road of Beijing in different perception dimensions, providing strong data support and a new spatial planning perspective for optimizing urban spatial layout and improving the quality of life of residents.