A deep learning project that leverages Convolutional Neural Networks (CNNs) to automatically detect Pneumonia from chest X-ray images. Our model is trained on a publicly available dataset and achieves an impressive 97.12% accuracy on the test set.
✅ Preprocessing and augmentation of X-ray images ✅ Convolutional Neural Network (CNN) for classification ✅ Achieved 97.12% accuracy on test dataset ✅ Training & evaluation pipeline included ✅ Works on Google Colab and local environments
- Chest X-ray images (Normal vs Pneumonia)
- Preprocessing applied: resizing, normalization, and data augmentation (flip, rotation, zoom, etc.)
- Balanced dataset for robust training
- Input Layer → Preprocessed chest X-ray images
- Convolutional + Pooling Layers → Feature extraction
- Fully Connected Layers → Classification
- Output Layer → Binary prediction:
Normal ✅orPneumonia ⚠️
- Accuracy:
97.12% - Evaluation Metrics: Loss curve & Accuracy curve (visualized in the notebook)
| Image | Prediction |
|---|---|
| Normal Chest X-ray | ✅ Normal |
| Pneumonia Chest X-ray |
- Open the notebook in Google Colab
- Upload dataset or mount Google Drive
- Run all cells to train & test the model
# Clone this repository
git clone https://github.com/Aan9758/pneumonia-detection.git
# Navigate to project folder
cd pneumonia-detection
# Install dependencies
pip install -r requirements.txt
# Run the notebook / script
jupyter notebook pneumonia_detection.ipynb- Python 🐍
- TensorFlow / Keras 🤖
- OpenCV 🔍
- Matplotlib / Seaborn 📊
- Deploy model as a web app / mobile app
- Use transfer learning with pretrained CNNs (ResNet, VGG16, EfficientNet)
- Expand dataset for higher generalization
Contributions are welcome! Feel free to fork this repo and submit a pull request.
This project is licensed under the MIT License – you’re free to use, modify, and distribute.