Upload 4 files

README.md

@@ -1,14 +1,69 @@
----
-title: Brain Tumor Classification
-emoji: 🏢
-colorFrom: red
-colorTo: gray
-sdk: gradio
-sdk_version: 6.0.0
-app_file: app.py
-pinned: false
-license: mit
-short_description: brain tumor MRI classification using EfficientNetB3
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Brain Tumor Classification with CNN and EfficientNetB3
+
+![Grad-CAM Example 1](./gradcam1.png)
+![Grad-CAM Example 2](./gradcam2.png)
+
+## Overview
+This project implements a deep learning pipeline for classifying brain tumor MRI images into four categories: **glioma**, **meningioma**, **notumor**, and **pituitary**. The model leverages transfer learning with EfficientNetB3 and includes interpretability with Grad-CAM visualizations.
+
+## Motivation
+Brain tumor classification from MRI images is a critical task for early diagnosis and treatment planning. Manual analysis is time-consuming and subjective. This project aims to:
+- Automate tumor classification with high accuracy
+- Provide interpretable results for clinicians using Grad-CAM
+- Demonstrate a reproducible, professional deep learning workflow
+
+## Dataset
+- The dataset is not included due to size. Download it using the link in `dataset_link.txt`.
+- Organize the data as:
+  - `Training/` (with subfolders for each class)
+  - `Testing/` (with subfolders for each class)
+
+## Project Structure
+- `model.ipynb`: Main notebook with all steps, explanations, and visualizations
+- `best_weights_balanced.h5`: Best model weights (saved automatically)
+- `dataset_link.txt`: Dataset download link
+- `README.md`: This file
+- Grad-CAM example images: `gradcam1.png`, `gradcam2.png`
+
+## Approach & Steps
+1. **Data Loading & Visualization**
+   - Loads images using TensorFlow's `image_dataset_from_directory`
+   - Visualizes sample images and class distribution
+2. **Class Imbalance Analysis**
+   - Plots class counts to check for imbalance
+   - (If needed, you can add class weighting or augmentation)
+3. **Model Architecture**
+   - Uses EfficientNetB3 (pretrained on ImageNet) as feature extractor
+   - Adds custom dense, batch normalization, and dropout layers
+   - Output layer: softmax for 4 classes
+4. **Training**
+   - Early stopping, learning rate scheduling, and best weight saving
+   - Plots training/validation loss and accuracy
+5. **Evaluation**
+   - Prints test loss, accuracy, precision, recall, F1-score
+   - Shows confusion matrix
+6. **Interpretability: Grad-CAM**
+   - Generates Grad-CAM heatmaps for test images
+   - Helps understand which regions the model focuses on for its decision
+
+## Example Results
+Below are Grad-CAM visualizations showing the model's attention on MRI images:
+
+![Grad-CAM Example 1](./gradcam1.png)
+![Grad-CAM Example 2](./gradcam2.png)
+
+## How to Run
+1. Install requirements (see notebook for package list)
+2. Download and extract the dataset as described above
+3. Run `model.ipynb` step by step
+4. Review the outputs, metrics, and Grad-CAM visualizations
+
+## Key Points & Best Practices
+- **Transfer Learning**: EfficientNetB3 enables strong feature extraction with limited data
+- **Callbacks**: Early stopping and best weight saving prevent overfitting
+- **Visualization**: Data and training curves are visualized for transparency
+- **Interpretability**: Grad-CAM provides insight into model decisions
+- **Reproducibility**: All steps are documented and reproducible
+
+## License
+This project is for educational and research purposes only.

app.py

@@ -0,0 +1,87 @@
+import gradio as gr
+import numpy as np
+import tensorflow as tf
+from tensorflow.keras.models import load_model
+from tensorflow.keras.preprocessing import image
+import cv2
+
+# Model yükle
+model = load_model("best_weights_balanced.h5")
+
+# Sınıf isimleri
+class_names = ["Glioma", "Meningioma", "No Tumor", "Pituitary"]
+
+# Grad-CAM fonksiyonu
+def get_gradcam(img_array, model, last_conv_layer_name="top_conv"):
+    grad_model = tf.keras.models.Model(
+        [model.inputs],
+        [model.get_layer(last_conv_layer_name).output, model.output]
+    )
+    
+    with tf.GradientTape() as tape:
+        conv_outputs, predictions = grad_model(img_array)
+        pred_index = tf.argmax(predictions[0])
+        class_channel = predictions[:, pred_index]
+    
+    grads = tape.gradient(class_channel, conv_outputs)
+    pooled_grads = tf.reduce_mean(grads, axis=(0, 1, 2))
+    
+    conv_outputs = conv_outputs[0]
+    heatmap = conv_outputs @ pooled_grads[..., tf.newaxis]
+    heatmap = tf.squeeze(heatmap)
+    heatmap = tf.maximum(heatmap, 0) / tf.math.reduce_max(heatmap)
+    
+    return heatmap.numpy(), pred_index.numpy()
+
+def predict_and_explain(img):
+    # Görüntüyü hazırla
+    img_resized = cv2.resize(img, (300, 300))
+    img_array = np.expand_dims(img_resized / 255.0, axis=0)
+    
+    # Tahmin
+    predictions = model.predict(img_array, verbose=0)
+    pred_class = np.argmax(predictions[0])
+    confidence = predictions[0][pred_class]
+    
+    # Grad-CAM
+    heatmap, _ = get_gradcam(img_array, model)
+    heatmap = cv2.resize(heatmap, (img_resized.shape[1], img_resized.shape[0]))
+    heatmap = np.uint8(255 * heatmap)
+    heatmap_colored = cv2.applyColorMap(heatmap, cv2.COLORMAP_JET)
+    
+    # Overlay
+    superimposed = cv2.addWeighted(img_resized, 0.6, heatmap_colored, 0.4, 0)
+    
+    # Sonuçlar
+    results = {class_names[i]: float(predictions[0][i]) for i in range(4)}
+    
+    return results, superimposed
+
+# Gradio arayüzü
+demo = gr.Interface(
+    fn=predict_and_explain,
+    inputs=gr.Image(label="MRI Görüntüsü Yükle"),
+    outputs=[
+        gr.Label(num_top_classes=4, label="Tahmin"),
+        gr.Image(label="Grad-CAM Açıklaması")
+    ],
+    title="🧠 Beyin Tümörü MRI Sınıflandırma",
+    description="""
+    **EfficientNetB3 + Grad-CAM ile Açıklanabilir AI**
+    
+    Bu model beyin MRI görüntülerini 4 kategoride sınıflandırır:
+    - **Glioma** - Glial hücrelerden kaynaklanan tümör
+    - **Meningioma** - Meninks zarından kaynaklanan tümör  
+    - **Pituitary** - Hipofiz bezi tümörü
+    - **No Tumor** - Tümör yok
+    
+    Grad-CAM, modelin hangi bölgelere odaklandığını gösterir.
+    
+    ⚠️ *Bu araç sadece eğitim amaçlıdır, tıbbi teşhis için kullanılamaz.*
+    """,
+    examples=[],  # Örnek görüntü ekleyebilirsin
+    theme=gr.themes.Soft()
+)
+
+if __name__ == "__main__":
+    demo.launch()

best_weights_balanced.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc5f885e409ea0e1db1328870676557141420d4f4629f4f7b1a980177092507f
+size 45567592

requirements.txt

@@ -0,0 +1,5 @@
+gradio==4.44.0
+tensorflow==2.15.0
+numpy
+opencv-python-headless
+Pillow