X-ray Pneumonia Classification using TPUs
Este es mi portafolio de proyectos.
View the Project on GitHub alxmares/X-ray_Pneumonia_Classification
Tools and Technologies Used
Key Results
These results indicate a high performance of the model in distinguishing between normal and pneumonia-affected X-ray images.
Model Architecture
- Base Model: Utilized a pre-trained InceptionV3 model on ImageNet, which was fine-tuned on the X-ray dataset to improve its sensitivity and specificity for pneumonia classification.
- Layers Added: Custom dense layers added to adapt the model to the binary classification task.
- Optimizer: Adam optimizer with a learning rate scheduler to ensure efficient training.
- Loss Function: Binary cross-entropy to handle the two-class problem.
Dataset
- Source: The dataset used in this project is from the Chest X-Ray Images (Pneumonia), which is a publicly available dataset consisting of 5,863 X-ray images.
- Classes: The dataset contains two classes: “NORMAL” and “PNEUMONIA”.
- Preprocessing: Images were resized to 180x180 pixels, and data augmentation was applied to improve model generalization.
