• Sai Krishna K
• Sasikala R

Corona Virus is an unexpected catastrophe that led to a devastating effect on human health. This paper aims to find the exact location of the lesion for estimating the disease progression using the Deep Learning technique like CNN as it is proven to be the best for the medical images. In this paper to enhance the quality and performance of the dataset in the pre-processing certain techniques are used namely CLAHE, Data Augmentation, cropping the images using the contour technique, and removal of images with no corresponding mask image. Later in this paper, a comparative analysis is done using K fold Cross Validation between different models like UNet segmentation, UNet segmentation with pre-trained InceptionResNetV2 as its backbone, UNet segmentation with pre-trained DenseNet201 as its backbone, and UNet ++ segmentation to find out the best model in identifying the exact location of the region of infection and for the best utilization of resources in this paper, Adam’s Optimization is used with Binary Cross Entropy Dice loss function. After the experimental analysis, UNet segmentation with pre-trained DenseNet201 as its backbone got the best Dice Similarity Coefficient of 95.883

Lesion segmentation, Transfer Learning, UUNet, UNet++, DenseNet201, and InceptionResNetV2

"Deep Learning models for Covid 19 Lesion Identification using CT Lung Images", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.9, Issue 6, page no.80-89, June-2022, Available :http://www.jetir.org/papers/JETIRFM06015.pdf

"Deep Learning models for Covid 19 Lesion Identification using CT Lung Images", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.9, Issue 6, page no. pp80-89, June-2022, Available at : http://www.jetir.org/papers/JETIRFM06015.pdf