Salvato in:
Dettagli Bibliografici
Autori principali: Vipulananthan, Vipooshan, Mohottala, Kumudu, Chinthana, Kavindu, Paramulla, Nimsara, Chitraranjan, Charith D
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2508.15216
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909976573771776
author Vipulananthan, Vipooshan
Mohottala, Kumudu
Chinthana, Kavindu
Paramulla, Nimsara
Chitraranjan, Charith D
author_facet Vipulananthan, Vipooshan
Mohottala, Kumudu
Chinthana, Kavindu
Paramulla, Nimsara
Chitraranjan, Charith D
contents Accident prediction and timely preventive actions improve road safety by reducing the risk of injury to road users and minimizing property damage. Hence, they are critical components of advanced driver assistance systems (ADAS) and autonomous vehicles. While many existing systems depend on multiple sensors such as LiDAR, radar, and GPS, relying solely on dash-cam videos presents a more challenging, yet more cost-effective and easily deployable solution. In this work, we incorporate improved spatio-temporal features and aggregate them through a recurrent network to enhance state-of-the-art graph neural networks for predicting accidents from dash-cam videos. Experiments using three publicly available datasets (DAD, DoTA and DADA) show that our proposed STAGNet model achieves higher average precision and mean time-to-accident scores than previous methods, both when cross-validated on a given dataset and when trained and tested on different datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2508_15216
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle STAGNet: A Spatio-Temporal Graph and LSTM Framework for Accident Anticipation
Vipulananthan, Vipooshan
Mohottala, Kumudu
Chinthana, Kavindu
Paramulla, Nimsara
Chitraranjan, Charith D
Computer Vision and Pattern Recognition
Accident prediction and timely preventive actions improve road safety by reducing the risk of injury to road users and minimizing property damage. Hence, they are critical components of advanced driver assistance systems (ADAS) and autonomous vehicles. While many existing systems depend on multiple sensors such as LiDAR, radar, and GPS, relying solely on dash-cam videos presents a more challenging, yet more cost-effective and easily deployable solution. In this work, we incorporate improved spatio-temporal features and aggregate them through a recurrent network to enhance state-of-the-art graph neural networks for predicting accidents from dash-cam videos. Experiments using three publicly available datasets (DAD, DoTA and DADA) show that our proposed STAGNet model achieves higher average precision and mean time-to-accident scores than previous methods, both when cross-validated on a given dataset and when trained and tested on different datasets.
title STAGNet: A Spatio-Temporal Graph and LSTM Framework for Accident Anticipation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.15216