Proceedings of the First International Conference on Computing, Communication and Control System, I3CAC 2021, 7-8 June 2021, Bharath University, Chennai, India

Research Article

Classification Of Covid19 Using Deep Neural Network

Download594 downloads
  • @INPROCEEDINGS{10.4108/eai.7-6-2021.2308563,
        author={Thaiyalnayaki  K.},
        title={Classification Of Covid19 Using Deep Neural Network},
        proceedings={Proceedings of the First International Conference on Computing, Communication and Control System, I3CAC 2021, 7-8 June 2021, Bharath University, Chennai, India},
        publisher={EAI},
        proceedings_a={I3CAC},
        year={2021},
        month={6},
        keywords={covid 19 deep cnn x ray automatic classifier},
        doi={10.4108/eai.7-6-2021.2308563}
    }
    
  • Thaiyalnayaki K.
    Year: 2021
    Classification Of Covid19 Using Deep Neural Network
    I3CAC
    EAI
    DOI: 10.4108/eai.7-6-2021.2308563
Thaiyalnayaki K.1,*
  • 1: Department of Electronics and Communication Engineering, SRM Institute of science and Technology, Ramapuram
*Contact email: thaiyalk@srmist.edu.in

Abstract

The ongoing logical researches are applying AI and machine learning in the field of clinical pneumonia and Covid in X-ray beam arrangement. A Deep CNN based model can distinguish the COVID-19 cases at a quicker speed by identifying the features of infected patients. The trainset consists of chest xrayof 80 healthy people and 80 covid people.The test set include 20 normal people and 20 covid affected chest xrays.Keras sequential Deep CNN classifier is constructed with convolution1,pooling1, convolution2,pooling2, flattening and dense fully connected net.The input shape of 2D convolution layer is of size 64x64x3 with a 3x3 mask and ReLu activation function. Max pooling mask size is 2x2.Conolution layer2 and max pool layer2 utilize the mask 3x3 and 2x2.After straightening, the completely connected layer comprises of 128 neurons with Relu activation function and one sigmoid neuron.Binary cross entropy loss function and Adam optimizer are used in CNN. The fully connected output layer consists of 129 neurons , out of which 128 are relu activation with 8,02,944 trainable parameters with accuracy obtained is 97%, which can be improved by increasing the dataset.This model act as an automatic classification of covid 19 and normal subjects.