Open Collections

Elbishlawi, Sherif

University of British Columbia

The COVID-19 pandemic has imposed significant challenges on countries all over the world. The main goal of this thesis is to investigate the use of computer vision and deep learning application in the fight against the COVID-19 pandemic. To achieve this goal, a survey on the current state-of-the-art methods in crowd analysis, including crowd counting and crowd action recognition is first performed. This survey is the first published comprehensive survey that covers both traditional and deep learning approaches for crowd counting and crowd action recognition. Based on the outcome of the survey and to help enforce the current World Health Organization (WHO) recommendations for maintaining a social distance of 6 ft between people in crowded areas, we propose SocialNet, a novel deep learning computer vision-based algorithm that can be used to actively monitor video streams from public surveillance cameras and detect violations. SocialNet is a novel network that uses DETR as an object detector along with an autoencoder-decoder network design for detecting social distancing violations in crowd scene. SocialNet was tested on the Oxford Town Centre dataset and achieved an accuracy of 95.47%. Finally, we apply computer vision and deep learning in the area of detecting COVID-19 from chest X-ray images. We propose CORONA-Net, a novel algorithm based on re-initialization and classification of abnormal X-ray chest infections. CORONA-Net uses an Encoder-Decoder network structure to diagnose COVID-19 from chest X-ray images. Extensive experiments were performed on a publicly available dataset, COVIDx, and the results show that CORONA-Net significantly outperforms the current state-of-the-art networks with an overall accuracy of 95.84%.

Text

2021-09-15

Attribution-ShareAlike 4.0 International

http://hdl.handle.net/2429/79714

Computer Science

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-sa/4.0/