/MSc-Project-Part1-Pneumonia-detection-using-tansfer-learning-on-ResNet50-VGG16-and-VGG19

The dataset having Pneumonia and Normal chest X-Ray images were trained on different numbers of epochs to check the variability in the training and validation accuracies. The ResNet50 model with the highest and closest Training and Validation accuracies was then used for the prediction.

Primary LanguageJupyter Notebook

Pneumonia-detection-using-tansfer-learning-on-ResNet50, VGG16 and VGG19.

  1. This is the part of my MSC project which is solving the problem of classifying the healthy and pneumonia chest x-ray images using CNN.
  2. The main aim was to check the variability in the training and testing accuracies of the individual models (ResNet50, VGG16 and VGG19) on different number of epochs and use the model with highest and closest training and testing accuracies for the prediction.
  3. Comparing the highest and closest training and testing accuracies obtained from the 3 convolutional neural networks, using the best model of three model for prediction.
  4. Dataset containing 5863 Chest X-Ray images of Pneumonia and Healthy lungs were trained on RESNET50, VGG16 and VGG19 to check the variability in the training and testing accuracies.
  5. Whole dataset was trained on different number of epochs (20, 25, 20, 5) 4 times individually to observe the performance of the mentioned CNN models.
  6. In total 12 models were trained individually. Below are the results obtained: image image image