Open AccessCongestion Control early warning system using Deep Learning
Author(s) -
N Sandeep,
Ragul N.S,
Nikil Dhas P,
V Vaishnavi
Publication year - 2021
Publication title -
international journal of computer communication and informatics
Language(s) - English
Resource type - Journals
ISSN - 2582-2713
DOI - 10.34256/ijcci2124
Subject(s) - computer science , convolutional neural network , upload , artificial intelligence , cloud computing , real time computing , network congestion , matlab , video tracking , software , crowding , object (grammar) , computer vision , data mining , computer network , operating system , neuroscience , network packet , biology
A new approach is proposed to analyze the live crowd and to provide an alert at the time of congestion, over-crowding and sudden gathering of pedestrians in a particular region. This paper proposes a completely software-oriented approach using MATLAB where it uses object detection and object tracking using Faster R- CNN (Region Based Convolutional Neural Network) algorithm where inception model of Google is used as CNN model which is pre-trained. This proposed method gives significant result on proposed dataset and the crowd congestion using Faster R-CNN approach which gives an accuracy of 93.503% at the rate 28 frames per second and the crowd detected video frames are uploaded to cloud storage.