Update app.py

app.py

@@ -1,46 +1,39 @@
-import tensorflow as tf
 import gradio as gr
-from tensorflow.keras.models import load_model
 import numpy as np
-from tensorflow.keras.preprocessing import image
+from keras.models import load_model
+from PIL import Image
 
-# ---- Define custom metrics ----
-def recall(y_true, y_pred):
-    true_positives = tf.keras.backend.sum(tf.keras.backend.round(tf.keras.backend.clip(y_true * y_pred, 0, 1)))
-    possible_positives = tf.keras.backend.sum(tf.keras.backend.round(tf.keras.backend.clip(y_true, 0, 1)))
-    recall = true_positives / (possible_positives + tf.keras.backend.epsilon())
-    return recall
+# Load the trained model (no custom_objects needed if using built-in metrics only)
+model = load_model("pneumonia_cnn_model.h5")
 
-def precision(y_true, y_pred):
-    true_positives = tf.keras.backend.sum(tf.keras.backend.round(tf.keras.backend.clip(y_true * y_pred, 0, 1)))
-    predicted_positives = tf.keras.backend.sum(tf.keras.backend.round(tf.keras.backend.clip(y_pred, 0, 1)))
-    precision = true_positives / (predicted_positives + tf.keras.backend.epsilon())
-    return precision
+# Preprocessing function for uploaded X-ray images
+def preprocess_image(image):
+    image = image.convert("L")  # Convert to grayscale
+    image = image.resize((150, 150))  # Resize to model's expected input
+    image_array = np.array(image) / 255.0  # Normalize pixel values
+    image_array = np.expand_dims(image_array, axis=-1)  # Add channel dimension
+    image_array = np.expand_dims(image_array, axis=0)  # Add batch dimension
+    return image_array
 
-def f1(y_true, y_pred):
-    p = precision(y_true, y_pred)
-    r = recall(y_true, y_pred)
-    return 2 * ((p * r) / (p + r + tf.keras.backend.epsilon()))
-
-# ---- Load the model ----
-model = load_model("pneumonia_cnn_model.h5", custom_objects={'f1': f1, 'precision': precision, 'recall': recall})
-
-# ---- Define prediction function ----
-def predict_image(img):
-    img = img.resize((150, 150))  # Make sure this matches your training image size
-    img = image.img_to_array(img)
-    img = np.expand_dims(img, axis=0)
-    img = img / 255.0
-    prediction = model.predict(img)[0][0]
+# Prediction function
+def predict(image):
+    image_array = preprocess_image(image)
+    prediction = model.predict(image_array)[0][0]
     label = "Pneumonia" if prediction > 0.5 else "Normal"
-    return f"Prediction: {label} ({prediction:.2f})"
+    confidence = prediction if prediction > 0.5 else 1 - prediction
+    return f"{label} ({confidence * 100:.2f}% confidence)"
 
-# ---- Create Gradio Interface ----
-interface = gr.Interface(fn=predict_image,
-                         inputs=gr.Image(type="pil"),
-                         outputs="text",
-                         title="Pneumonia Detection",
-                         description="Upload a chest X-ray image to predict if it shows signs of pneumonia.")
+# Gradio interface
+interface = gr.Interface(
+    fn=predict,
+    inputs=gr.Image(type="pil"),
+    outputs="text",
+    title="Pneumonia Detection from Chest X-rays",
+    description="Upload a chest X-ray image to detect if the person has pneumonia.",
+    theme="huggingface",
+    allow_flagging="never"
+)
 
-# ---- Launch ----
-interface.launch()
+# Launch the app
+if __name__ == "__main__":
+    interface.launch()