We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
An IoT Enabled Vehicular Decision Fusion Framework for Accident Detection and Classification.
- Authors
Kumar, Nikhil; Acharya, Debopam; Lohani, Divya
- Abstract
Increased number of vehicle-based road accidents is a key reason for the death and disability of people. Timely information on accidents can save lives. Current accident detection systems are either working towards increasing the accuracy of detection or the severity of the accident. Accurate information of an accident type can help the emergency medical services (EMS) to identify the most appropriate type of rescue and medical assistance to the victims. This work introduces a smartphone-based accident detection and classification (ADC) system that not only detects the accident but also classifies the type of accident as collision, rollover, or fall-off, using internal and external sensors. We have developed an end-to-end IoT system that exploits a multi-sensor data fusion framework to accurately classify the type of accident. The framework combines the decisions of three different classifiers based on Nave Bayes (NB), K-Nearest Neighbor (KNN), and Random Forest (RF) methods using stacking ensemble approach. Logistic Regression based stacking approach is found to be highly accurate in comparison to NB, KNN, and RF classifiers when they were used individually.
- Publication
International Journal of Next-Generation Computing, 2020, Vol 11, Issue 2, p191
- ISSN
2229-4678
- Publication type
Academic Journal