This is a case study comparing Resnet18 and Vgg16 models for pneumonia detection from chest X-rays. The study includes exploratory data analysis (EDA) and model comparison.
- Both VGG16 and Resnet18 is implemented from scratch and using pretrained model from PyTorch.
- If using pretrained model, make sure to enable internet on kaggle.
To obtain the datasets necessary for this case study, please download the following Kaggle datasets:
Dataset: https://www.kaggle.com/datasets/paultimothymooney/chest-xray-pneumonia
- EDA Dataset: https://www.kaggle.com/nischallal/eda-detection-of-pneumonia-in-x-ray
- Resnet18 Model from Scratch: https://www.kaggle.com/code/nischallal/resnet18-detection-of-pneumonia-in-x-ray
- Vgg16 Model from Scratch: https://www.kaggle.com/code/nischallal/vgg16-scratch-detection-of-pneumonia-in-x-ray
To reproduce the results or conduct further research, follow these steps:
- Run the notebooks from link.
- Download the datasets from the provided Kaggle links.
- Run the provided Jupyter Notebooks to perform EDA and compare the models.