|
|
--- |
|
|
license: mit |
|
|
metrics: |
|
|
- accuracy |
|
|
library_name: keras |
|
|
--- |
|
|
|
|
|
# Malaria Detection - Custom CNN Model |
|
|
|
|
|
## Model Description |
|
|
|
|
|
This is a custom Convolutional Neural Network (CNN) trained to detect malaria parasites in cell images. The model classifies blood cell images as either **Parasitized** or **Uninfected**. |
|
|
|
|
|
## Model Architecture |
|
|
|
|
|
- 3 Convolutional layers with ReLU activation |
|
|
- MaxPooling layers for downsampling |
|
|
- Dropout layer (0.5) for regularization |
|
|
- Dense layers for classification |
|
|
- Binary output with sigmoid activation |
|
|
|
|
|
## Training Details |
|
|
|
|
|
- **Dataset**: Cell Images for Detecting Malaria |
|
|
- **Input Size**: 150x150x3 |
|
|
- **Optimizer**: Adam |
|
|
- **Loss Function**: Binary Crossentropy |
|
|
- **Epochs**: 10 |
|
|
- **Validation Split**: 20% |
|
|
|
|
|
## Performance |
|
|
|
|
|
- **Validation Accuracy**: {history.history['val_accuracy'][-1]:.4f} |
|
|
- **Validation Loss**: {history.history['val_loss'][-1]:.4f} |
