/covid-chest-xray-classifier-classical

Primary LanguageJupyter Notebook

covid-chest-xray-classifier-classical

Problem statement -

Detecting Covid19 or Viral Pneumonia using Chest x-ray

Dataset used -

I have used this datset from Kaggle which contains 250 training and 65 testing images for our model.

alt text

Plotting accuracy, cost, adaptive learning rate -

Accuracy

alt text

Training and Validation Loss

alt text

Adaptive Learning Rate

alt text

Predicting on individual images -

This model has 87.5% validation accuracy as tested on the given test images. Although there is still some room for improvement using larger dataset for both training and testing.

Correct Predictions

Label: Covid , Predicted: Covid

alt text

Label: Covid , Predicted: Covid

alt text

Label: Normal , Predicted: Normal

alt text

Label: Normal , Predicted: Normal

alt text

Label: Viral Pneumonia , Predicted: Viral Pneumonia

alt text

Label: Viral Pneumonia , Predicted: Viral Pneumonia

alt text

Wrong Predictions

Label: Covid , Predicted: Normal

alt text

Disclaimer -

87.5% accuracy is something really huge to claim and we can't do that exactly here. One of the reasons is that the dataset is not enough large to be trained on for a industry level deep neural network architecture.

So there is a scope of future improvement by training on more data