This repository contains a deep learning project focused on detecting pneumonia from chest X-ray images. The project utilizes convolutional neural networks (CNNs) for image classification, aiming to differentiate between normal and pneumonia-infected lungs.
Pneumonia is a serious lung infection that requires timely diagnosis and treatment. Automated detection using deep learning techniques can assist in improving diagnostic accuracy and speed. This project implements a deep learning pipeline to classify chest X-rays into two categories: Normal and Pneumonia.
Pneumonia_Detection.ipynb: The Jupyter Notebook containing the full pipeline, from data loading and preprocessing to model training and evaluation.data/: Directory expected to contain the training, validation, and test images organized in subfolders (e.g.,train/PNEUMONIA/,train/NORMAL/, etc.).models/: Directory to save trained models and weights (not included by default).
- Data Preprocessing: Images are resized to 224x224 pixels and normalized. Data augmentation techniques are applied to improve model generalization.
- Model Architecture: A CNN model built using Keras, leveraging transfer learning with the VGG16 architecture.
- Evaluation Metrics: The model's performance is evaluated using accuracy, precision, recall, F1 score, and ROC-AUC.
- TensorFlow
- Keras
- OpenCV
- Scikit-learn
- Matplotlib
- Seaborn