Create README.md

README.md

@@ -0,0 +1,105 @@
+---
+library_name: keras
+tags:
+- medical
+- medical-imaging
+- chest-xray
+- xray
+- covid-19
+---
+# Inception V3 Covid-19 Classifier
+
+## Model description
+
+This model is a fine-tuned Inception V3 convolutional neural network designed for chest X-ray image classification, with a focus on Covid-19 detection.
+
+The model was trained using transfer learning from ImageNet weights and further fine-tuned on medical imaging data.
+
+An optionnal lung auto-masking preprocessing step can be applied upstream to reduce background bias and improve interpretability.
+
+This model is entended for research and educational purposes only.
+
+## Intended use
+
+- Binary or multi-class classification of chest X-ray images
+- Research on medical image classification
+- Demonstration of deep learning pipelines (preprocessing, fine-tuning, Grad-CAM)
+
+## Not intended for
+- Clinical diagnosis
+- Medical decision-making
+- Production healthcare systems
+
+## Training data
+
+The model was trained on the **COVID-19 Radiography Dataset (Kaggle)**:
+
+https://www.kaggle.com/datasets/tawsifurrahman/covid19-radiography-database/data
+
+The dataset contains chest X-ray images labeled as:
+- COVID
+- Normal
+- Lung-Opacity
+- Viral Pneumonia
+
+Only frontal chest X-ray images were considered.
+
+## Input format
+
+- Image size : 299x299 px
+- Channels : 3 (RBG)
+- Pixel values : normalized to [-1,1]
+
+## Output format
+
+- Class probabilites (softmax)
+- Final prediction via argmax
+
+## Training details
+
+- Base model : Inception V3 (ImageNet pretrained)
+- Fine-tuning : last 20 layers unfrozen
+- Optimizer : SGD
+- Loss function : Categorical crossentropy
+
+## Evaluation results
+
+The model achieves an accuracy of approximately **84%** on the validation set.
+
+Performance metrics were evaluated using :
+- Accuracy
+- Confusion matrix
+- ROC/AUC curves
+
+Results are dataset-dependent and may vary under different data distributions.
+
+## Explainability
+
+Model predictions can be interpreted using Grad-CAM visualizations, highlighting salient lung regions contributing to the decision.
+
+## Limitations
+
+- Trained on a single public dataset
+- Sensitive to image quality and acquisition protocol
+- Performance may degrade on non-frontal or low-quality X-ray
+- Not validated on clinical-grade datasets
+
+## Ethical considerations
+
+This model is provided for research purposes only.
+It must not be used as a diagnostic tool or to guide clinical decisions.
+
+## Example usage
+
+```python
+from tensorflow.keras.models import load_model
+import numpy as np
+
+model = load_model("inception_V3_covid.keras")
+
+preds = model.predict(input_tensor)
+```
+
+## Author
+
+Asma Sima