A deep learning model built using TensorFlow/Keras to detect bone fractures in X-ray images. The model uses a modified Xception architecture with custom layers for binary classification of fracture/non-fracture cases.
A deep learning model built using TensorFlow/Keras to detect bone fractures in X-ray images. The model uses a modified Xception architecture with custom layers for binary classification of fracture/non-fracture cases. Well I tried resnet, efficientnet and inception architecture but the model didn't performed well.
- Binary classification of X-ray images (fracture/normal)
- Data preprocessing and augmentation
- Balanced dataset handling
- Transfer learning using Xception architecture
- Custom model architecture with regularization
This project uses X-ray images dataset on kaggle which can be accessed through:
- bone fracture detection using x-rays: https://www.kaggle.com/datasets/vuppalaadithyasairam/bone-fracture-detection-using-xrays/data
- Size: 183 MB
- Classes: Fractured and Non-fractured X-rays
Base Model: Xception (pre-trained on ImageNet) Additional layers:
GaussianNoise (0.2) GlobalAveragePooling2D Dense layers (256, 128 units) Batch Normalization Dropout (0.3)
Training Accuracy: 96.2% Validation Accuracy: 98.5%
python tensorflow==2.x numpy pandas scikit-learn imbalanced-learn pillow opencv-python
Future Improvements
- Add data augmentation techniques
- Implement cross-validation
- Add visualization of results
- Deploy model as web application