Adaptive Context Based Road Accident Risk Prediction Using Spatio-Temporal Deep Learning | IEEE Journals & Magazine | IEEE Xplore

Adaptive Context Based Road Accident Risk Prediction Using Spatio-Temporal Deep Learning


Impact Statement:According to the World Health Organization (WHO), road accidents kill 1.3 million people and injure 50 million people each year, resulting in a global cost of US $518 bil...Show More

Abstract:

Traffic accidents are common urban events that pose significant risks to human safety, traffic management, and economic stability; consequently, the research community is...Show More
Impact Statement:
According to the World Health Organization (WHO), road accidents kill 1.3 million people and injure 50 million people each year, resulting in a global cost of US $518 billion (World Health Organization, 2018) [1]. Therefore, accident prediction would assist the government in laying down safety measures and traffic planning in risk-prone areas. Accident prediction, however, is a heterogeneous spatiotemporal deep learning task that involves location, time, and contextual factors, for which existing convolution or recurrent neural networks are difficult to utilize because they are not designed for spatiotemporal data specifically. Moreover, existing accident prediction methods cannot simultaneously account for sparsity and multiple contextual factors, which have a significant impact on the performance of proposed solutions. This work aims to improve the performance of accident prediction by combining graph and deep learning solutions to reduce error approximately by 4% (RMSE) and increase...

Abstract:

Traffic accidents are common urban events that pose significant risks to human safety, traffic management, and economic stability; consequently, the research community is paying increasing attention toward accident risk prediction. However, accident risk prediction is a challenging problem because accident occurrences are sparse and influenced by multiple contextual factors (e.g., POI, road structure, road type, hour of the day, and month). Therefore, in this article, we propose a novel architecture named Topographic-Weighted Context Category (TWCCnet) that adapts heterogeneous contextual category weights based on spatial–temporal correlations across sectors. Specifically, the framework consists of two parallel components: one uses convolution and stacked bidirectional gated recurrent unit (Bi-GRU) to capture spatial–temporal relationships between neighborhood sectors, while the other uses multiple graph convolution network (GCN) over resemblance graphs to capture spatial–temporal rela...
Published in: IEEE Transactions on Artificial Intelligence ( Volume: 5, Issue: 6, June 2024)
Page(s): 2872 - 2883
Date of Publication: 03 November 2023
Electronic ISSN: 2691-4581

Funding Agency:


References

References is not available for this document.