Scholarlink

<p>Effective models for analyzing and predicting pedestrian flow are important for ensuring the safety of both pedestrians and other road users. These models also play a crucial role in optimizing infrastructure design, improving traffic management, and supporting the economic utility of interconnected communities. The deployment of city-wide automatic pedestrian counting systems provides researchers with invaluable data, enabling the development and training of deep learning applications that offer better insights into traffic patterns and crowd dynamics. Benefiting from real-world data provided by the City of Melbourne and City of Casey pedestrian counting systems, this study presents a pedestrian flow prediction model, an extension of the Diffusion Convolutional Grated Recurrent Unit (DCGRU) incorporating dynamic time warping, named DCGRU-DTW. This model has the spatial dependencies of pedestrian flow captured through the diffusion convolution operation and the temporal dependency captured by the gated recurrent unit. Through extensive numerical experiments, we demonstrate that the proposed model outperforms the classic vector autoregressive model and the original DCGRU across multiple model accuracy metrics.</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

Pedestrian Volume Prediction Using a Diffusion Convolutional Gated Recurrent Unit Model with Dynamic Time Warping

  • 期刊论文

  • 发表日期：

    2025年6月9日

查看全文
  • Yiwei Dong,
  • Tingjin Chu,
  • Lele Zhang,
  • Hadi Ghaderi,
  • Hanfang Yang

摘要

Effective models for analyzing and predicting pedestrian flow are important for ensuring the safety of both pedestrians and other road users. These models also play a crucial role in optimizing infrastructure design, improving traffic management, and supporting the economic utility of interconnected communities. The deployment of city-wide automatic pedestrian counting systems provides researchers with invaluable data, enabling the development and training of deep learning applications that offer better insights into traffic patterns and crowd dynamics. Benefiting from real-world data provided by the City of Melbourne and City of Casey pedestrian counting systems, this study presents a pedestrian flow prediction model, an extension of the Diffusion Convolutional Grated Recurrent Unit (DCGRU) incorporating dynamic time warping, named DCGRU-DTW. This model has the spatial dependencies of pedestrian flow captured through the diffusion convolution operation and the temporal dependency captured by the gated recurrent unit. Through extensive numerical experiments, we demonstrate that the proposed model outperforms the classic vector autoregressive model and the original DCGRU across multiple model accuracy metrics.