Update app.py

app.py

@@ -2,48 +2,40 @@ import gradio as gr
 import numpy as np
 from PIL import Image
 from keras.models import load_model
+import json
 
-# Load your pre-trained model (make sure the model file is in the same directory)
+# Load your pre-trained model
 model = load_model('brain_tumor_model.h5')
 
-# Function to process image and make predictions
-def predict_image(image):
-    # Resize the image to match model input
-    img = image.resize((128, 128))
-
-    # Convert the image to a NumPy array
-    img = np.array(img)
+# Load examples from JSON file
+with open('examples.json', 'r') as f:
+    examples_data = json.load(f)
 
-    # Ensure the image has 3 color channels (RGB)
-    if img.shape == (128, 128):  # If grayscale, convert to RGB
-        img = np.stack((img,) * 3, axis=-1)
+examples = [[example['image']] for example in examples_data]
 
-    # Normalize the image (if your model requires normalization)
-    img = img / 255.0
-
-    # Add a batch dimension (for model input)
-    img = np.expand_dims(img, axis=0)
+def predict_image(image):
+    try:
+        # Resize and preprocess the image
+        img = image.resize((128, 128))
+        img = np.array(img)
 
-    # Make the prediction
-    prediction = model.predict(img)
+        # Check and convert grayscale to RGB
+        if img.shape == (128, 128):  # Grayscale
+            img = np.stack((img,) * 3, axis=-1)
 
-    # Get the predicted class and confidence level
-    predicted_class = np.argmax(prediction)
-    confidence = np.max(prediction)
+        # Normalize the image
+        img = img / 255.0
+        img = np.expand_dims(img, axis=0)
 
-    # Return the result with confidence level
-    if predicted_class == 0:
-        return f'No tumor detected. Confidence: {confidence:.2f}'
-    else:
-        return f'Tumor detected. Confidence: {confidence:.2f}'
+        # Make the prediction
+        prediction = model.predict(img)
+        predicted_class = np.argmax(prediction)
+        confidence = np.max(prediction)
 
-# Provide example images (these should be paths to valid images)
-examples = [
-    ["https://huggingface.co/spaces/tmafantiri/braintumourdetector/blob/main/images/no/1%20no.jpeg"],  
-    ["https://huggingface.co/spaces/tmafantiri/braintumourdetector/blob/main/images/yes/Y1.jpg"],
-    ["https://huggingface.co/spaces/tmafantiri/braintumourdetector/blob/main/images/no/15%20no.jpg"],
-    ["https://huggingface.co/spaces/tmafantiri/braintumourdetector/blob/main/images/yes/Y104.jpg"]
-]
+        return f'{"No tumor detected" if predicted_class == 0 else "Tumor detected"}. Confidence: {confidence:.2f}'
+    
+    except Exception as e:
+        return f"Error: {str(e)}"
 
 # Create the Gradio interface
 iface = gr.Interface(
@@ -52,9 +44,9 @@ iface = gr.Interface(
     outputs=gr.Textbox(),
     title="Brain Tumor Detection AI App",
     description="Upload an image to detect brain tumors.",
-    theme="monochrome",  # Apply a dark theme to the interface
-    flagging_options=["Incorrect Diagnosis", "Image Not Clear", "Model Error"],  # Add flagging options
-    examples=examples  # Include example images
+    theme="monochrome",
+    flagging_options=["Incorrect Diagnosis", "Image Not Clear", "Model Error"],
+    examples=examples
 )
 
 # Launch the interface