This study presents an accident detection system leveraging the YOLOv3 model for real-time identification of head-on collisions, rear-end collisions, and vehicle rollovers. The YOLOv3 model was trained and fine-tuned on a unique dataset of accident photos following pre-training on the COCO dataset. Utilizing bounding box methodology, collisions were detected with high accuracy, achieving an average precision of 94% for vehicle rollovers, 92% for head-on collisions, and 93% for rear-end collisions. Remarkably the system performs in real-time, averaging 0.03 seconds for each frame of processing. By advancing technological innovation in accident detection systems, this research significantly contribute to the SDG 9. Moreover, its integration into intelligent transportation systems holds promise for enhancing infrastructure efficiency and safety. The proposed system represents a crucial step towards improving road safety, ultimately contributing to the advancement of sustainable infrastructure and innovation in transportation.
Skip Nav Destination
Article navigation
31 January 2025
INTERNATIONAL CONFERENCE ON ROBOTICS, AUTOMATION, AND INTELLIGENT COMPUTING: ICRAIC 2K24
18–19 May 2024
Pune, India
Research Article|
January 31 2025
Real time accident detection using bounding box methodology
Divya Uday Singh;
Divya Uday Singh
a)
Department of Computing Technologies, SRM Institute of Science and Technology
, Kattankulathur-603203, Chennai, India
Search for other works by this author on:
Piyush Singh;
Piyush Singh
b)
Department of Computing Technologies, SRM Institute of Science and Technology
, Kattankulathur-603203, Chennai, India
Search for other works by this author on:
Padmini Sankaramurthy
Padmini Sankaramurthy
c)
Department of Computing Technologies, SRM Institute of Science and Technology
, Kattankulathur-603203, Chennai, India
c)Corresponding author: [email protected]
Search for other works by this author on:
AIP Conf. Proc. 3255, 020008 (2025)
Citation
Divya Uday Singh, Piyush Singh, Padmini Sankaramurthy; Real time accident detection using bounding box methodology. AIP Conf. Proc. 31 January 2025; 3255 (1): 020008. https://doi.org/10.1063/5.0254418
Download citation file:
Pay-Per-View Access
$40.00
Sign In
You could not be signed in. Please check your credentials and make sure you have an active account and try again.
6
Views
Citing articles via
Inkjet- and flextrail-printing of silicon polymer-based inks for local passivating contacts
Zohreh Kiaee, Andreas Lösel, et al.
Design of a 100 MW solar power plant on wetland in Bangladesh
Apu Kowsar, Sumon Chandra Debnath, et al.
Effect of coupling agent type on the self-cleaning and anti-reflective behaviour of advance nanocoating for PV panels application
Taha Tareq Mohammed, Hadia Kadhim Judran, et al.
Related Content
Multi-camera person tracking and re-identification system
AIP Conf. Proc. (January 2025)