Accident Detection Using Fog Computing

Emergencies, in terms of definition, are the irregular and contiguous answer is a necessary demand for emergency administration. Worldwide, an important number of deaths arise each year, resulting in necessary delays in saving operations. To minimize these delays and save the lives of people when ha...

Full description

Saved in:
Bibliographic Details
Main Authors: Srivastava, Mayank, Dixit, Pratibha, Ranjan, Pulkit
Format: Conference Proceeding
Language:English
Subjects:
Accidental Detection
Collision avoidance
Computer networks
Delays
Fog Computing
Road transportation
Time factors
Training
Videos
Yolov3
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Emergencies, in terms of definition, are the irregular and contiguous answer is a necessary demand for emergency administration. Worldwide, an important number of deaths arise each year, resulting in necessary delays in saving operations. To minimize these delays and save the lives of people when have come up with a project that is based on the computer network that uses Fog computing. It contains three modules in which collision detection is to be carried out. Pre-recorded videos and live videos have been broken down into frames and sent to the MQTT broker which is a lightweight network protocol to send and receive messages. Video Frame is sent to Processing Unit for detection. Results are sent back to MQTT Broker and subscribed devices are informed accordingly. In this Live testing is being done on highways, apart from this Training of the model on various constraints using a mask and fast R-CNN is being done followed by a validation phase that shows the efficiency of this project. The result shows the need to save the lives which are in danger due to accidents by minimizing the time to contact the required department. This delay is minimized with the help of a concept called Fog Computing which serves as a middleware between devices and the cloud therefore saving time.
ISSN:2832-143X
DOI:10.1109/ISCON57294.2023.10111980