---
license: mit
language:
- en
metrics:
- accuracy
pipeline_tag: image-classification
---

## 📂 Project Structure

```
.
├── image_classifier.h5     # Trained model
├── main.py                 # Training & prediction script
└── README.md               # Project description
```

---

## ⚙️ Technologies Used

* Python 3.10+
* TensorFlow / Keras
* NumPy
* Matplotlib

---

## 📊 Dataset

The dataset is from **Kaggle Multi-Cancer Dataset**:

```
/kaggle/input/multi-cancer/Multi Cancer/Multi Cancer/Breast Cancer
```

Images are split into **90% training** and **10% validation** using `ImageDataGenerator`.

---

## 🏗️ Model Architecture

* **Conv2D (32 filters, 3x3, ReLU)**
* **MaxPooling2D (2x2)**
* **Conv2D (64 filters, 3x3, ReLU)**
* **MaxPooling2D (2x2)**
* **Conv2D (128 filters, 3x3, ReLU)**
* **MaxPooling2D (2x2)**
* **Flatten**
* **Dense (512, ReLU)**
* **Dense (Softmax output, # of classes)**

Optimizer: **Adam**
Loss: **Categorical Crossentropy**
Metric: **Accuracy**

---

## 🚀 Training

```python
model.fit(train_generator, validation_data=validation_generator, epochs=10)
```

After training, the model is saved as:

```python
model.save("image_classifier.h5")
```

---

## 🔮 Prediction Example

```python
def guess(image_path, model, class_indices):
    img = load_img(image_path, target_size=(150, 150))
    img_array = img_to_array(img) / 255.0
    img_array = np.expand_dims(img_array, axis=0)

    prediction = model.predict(img_array)
    predicted_class = np.argmax(prediction)
    class_labels = {v: k for k, v in class_indices.items()}
    predicted_label = class_labels[predicted_class]

    plt.imshow(img)
    plt.title(f"Model guess: {predicted_label}")
    plt.axis("off")
    plt.show()
```

---

## ✅ Results

* Trains a CNN model for breast cancer image classification
* Provides a simple **guess() function** to visualize predictions
* Model is reusable via `image_classifier.h5`