Update app.py

app.py

@@ -6,49 +6,60 @@ import cv2
 from keras.models import load_model
 from util import set_background  
 
-bg = cv2.imread('./bgs/bg5.png')
 
-blurred_bg = cv2.GaussianBlur(bg, (15, 15), 0)
+st.set_page_config(page_title="Pneumonia Classifier", layout="centered")
+st.title("🩺 Pneumonia Classifier Application")
+st.header("Upload a Chest X-ray Image")
 
-cv2.imwrite('./bgs/bg5_blur.png', blurred_bg)
 
-# Set background
-set_background('./bgs/bg5_blur.png')
+@st.cache_resource
+def load_bg():
+    bg_path = os.path.join(os.path.dirname(__file__), "bgs", "bg5.png")
+    bg = cv2.imread(bg_path)
+    if bg is None:
+        return None
+    blurred = cv2.GaussianBlur(bg, (15, 15), 0)
+    return blurred
 
-# Set title and header
-st.title('🩺 Pneumonia Classifier Application')
-st.header('Upload a Chest X-ray Image')
+@st.cache_resource
+def load_pneumonia_model():
+    model_path = os.path.join(
+        os.path.dirname(__file__),
+        "model",
+        "pneumonia_classifier.keras"
+    )
+    return load_model(model_path)
+
+
+bg_img = load_bg()
+if bg_img is not None:
+    tmp_bg = "/tmp/bg_blur.png"
+    cv2.imwrite(tmp_bg, bg_img)
+    set_background(tmp_bg)
 
-# Upload file
-file = st.file_uploader(
-    "Upload a chest X-ray image", 
-    type=['jpeg', 'jpg', 'png'],
-    label_visibility="visible"
-)
 
-# Load model
-model_path = os.path.join(os.path.dirname(__file__), 'model', 'pneumonia_classifier.keras')
-model = load_model(model_path)
+model = load_pneumonia_model()
+class_names = ["NORMAL", "PNEUMONIA"]
 
 
-class_names = ['NORMAL', 'PNEUMONIA']
+file = st.file_uploader(
+    "Upload a chest X-ray image",
+    type=["jpeg", "jpg", "png"]
+)
+
 
-# Display image and classify
 if file is not None:
-    image = Image.open(file).convert('L')  
+    image = Image.open(file).convert("L")
     st.image(image, caption="Uploaded X-ray", use_container_width=True)
 
-
     img = np.array(image)
     img = cv2.resize(img, (128, 128))
     img = img / 255.0
-    img = np.expand_dims(img, axis=(0, -1))  
+    img = np.expand_dims(img, axis=(0, -1))
 
-    # Predict
     prediction = model.predict(img)
     class_idx = np.argmax(prediction)
     confidence = np.max(prediction)
 
-    # Display results
-    st.write("## Prediction: {}".format(class_names[class_idx]))
-    st.write("### Confidence: {:.2f}%".format(confidence * 100))
+    st.success(f"Prediction: {class_names[class_idx]}")
+    st.info(f"Confidence: {confidence * 100:.2f}%")