This readme contains 2 portions: 1.COVID Detection through XRAY images 2.COVID Detection through XRAY images using focal loss.
This repository contains code and results for COVID-19 classification assignment by Deep Learning Spring 2020 course offered at Information Technology University, Lahore, Pakistan. This assignment is only for learning purposes and is not intended to be used for clinical purposes
VGG-16:
Accuracy of the network on the 1506 test images: 97 %
Accuracy of the network on the 1500 validation images: 89 %
Confusion Matrix for Test Set: [597 23 18 868]
Learning Rate = 0.001 Number of Layers added = 3 Linear() Relu() Dropout() Output() Batch Size = 16 No of threads used = 4 Momentum = 0.9 Loss = Cross Entropy and Foacal Loss Optimizer = SGD Epochs = 10
-------------------------------------------------
RESNET :
Accuracy of the network on the 1506 test images: 96 %
Accuracy of the network on the 1500 validation images: 88 %
Confusion Matrix for Test Set: [587 26 28 865]
Learning Rate = 0.001 Number of Layers added = 3 Linear() Relu() Dropout() Output() Batch Size = 16 No of threads used = 4 Momentum = 0.9 Loss = Cross Entropy Optimizer = SGD Epochs = 10
DATASET : https://drive.google.com/drive/u/1/folders/1rNAaDEfQbcGlumr_3Fewia4SObTrCHAk
WEIGHTS : https://drive.google.com/drive/u/1/folders/1YdsZGesIFWBIRM6P_TZ5TvLB6ie66nyN