Update app.py

app.py

@@ -9,35 +9,8 @@ from keras.layers import BatchNormalization, DepthwiseConv2D, TFSMLayer
 import os
 from io import BytesIO
 import base64
-st.markdown("""
-    <style>
-    .equal-cols {
-        display: flex;
-        gap: 1.5rem;
-    }
-    .equal-cols > div {
-        flex: 1;
-        display: flex;
-        flex-direction: column;
-    }
-    .lime-image {
-        width: 100%;
-        border-radius: 10px;
-    }
-    .overlay-box {
-        background-color: rgba(255, 255, 255, 0.85);
-        padding: 1rem;
-        border-radius: 10px;
-        overflow-y: auto;
-        color: #333;
-        font-size: 16px;
-        line-height: 1.5;
-        height: 100%;
-    }
-    </style>
-""", unsafe_allow_html=True)
 
-# --- Fix deserialization issues for BatchNorm and DepthwiseConv2D ---
+# --- Fix deserialization issues ---
 original_bn = BatchNormalization.from_config
 BatchNormalization.from_config = classmethod(
     lambda cls, config, *a, **k: original_bn(
@@ -49,7 +22,7 @@ DepthwiseConv2D.from_config = classmethod(
     lambda cls, config, *a, **k: original_dw({k: v for k, v in config.items() if k != "groups"}, *a, **k)
 )
 
-# --- Background Setup ---
+# --- Set Background ---
 def set_background(image_path):
     with open(image_path, "rb") as f:
         encoded = base64.b64encode(f.read()).decode()
@@ -81,14 +54,8 @@ set_background("5858.jpg")
 # --- Constants ---
 IMG_SIZE = (224, 224)
 CLASS_NAMES = [
-    'Normal',
-    'Diabetic Retinopathy',
-    'Glaucoma',
-    'Cataract',
-    'Age-related Macular Degeneration (AMD)',
-    'Hypertension',
-    'Myopia',
-    'Others'
+    'Normal', 'Diabetic Retinopathy', 'Glaucoma', 'Cataract',
+    'Age-related Macular Degeneration (AMD)', 'Hypertension', 'Myopia', 'Others'
 ]
 LIME_EXPLAINER = lime_image.LimeImageExplainer()
 
@@ -100,14 +67,25 @@ def load_model():
         st.error(f"Model folder '{model_path}' not found.")
         st.stop()
     try:
-        # Load TensorFlow SavedModel as an inference-only Keras layer
         model = tf.keras.Sequential([TFSMLayer(model_path, call_endpoint="serving_default")])
         return model
     except Exception as e:
         st.error(f"Error loading model: {e}")
         st.stop()
 
-# --- Preprocessing ---
+# --- Prediction ---
+def predict(images, model):
+    images = np.array(images)
+    preds = model.predict(images, verbose=0)
+    if isinstance(preds, dict):
+        for v in preds.values():
+            if isinstance(v, (np.ndarray, list)):
+                return np.array(v)
+        return np.array(list(preds.values())[0])
+    else:
+        return preds
+
+# --- Preprocessing Steps ---
 def preprocess_with_steps(img):
     h, w = img.shape[:2]
     center, radius = (w // 2, h // 2), min(w, h) // 2
@@ -115,22 +93,18 @@ def preprocess_with_steps(img):
     dist = np.sqrt((X - center[0]) ** 2 + (Y - center[1]) ** 2)
     mask = (dist <= radius).astype(np.uint8)
 
-    # Apply mask and replace black background with white
     circ = img.copy()
-    white_background = np.ones_like(circ, dtype=np.uint8) * 255  # White background
-    circ = np.where(mask[:, :, np.newaxis] == 1, circ, white_background)
+    white_bg = np.ones_like(circ, dtype=np.uint8) * 255
+    circ = np.where(mask[:, :, np.newaxis] == 1, circ, white_bg)
 
-    # CLAHE
     lab = cv2.cvtColor(circ, cv2.COLOR_RGB2LAB)
     cl = cv2.createCLAHE(clipLimit=2.0).apply(lab[:, :, 0])
     merged = cv2.merge((cl, lab[:, :, 1], lab[:, :, 2]))
     clahe_img = cv2.cvtColor(merged, cv2.COLOR_LAB2RGB)
 
-    # Sharpen + Resize
     sharp = cv2.addWeighted(clahe_img, 4, cv2.GaussianBlur(clahe_img, (0, 0), 10), -4, 128)
     resized = cv2.resize(sharp, IMG_SIZE) / 255.0
 
-    # Visualization
     fig, axs = plt.subplots(1, 4, figsize=(16, 4))
     for ax, image, title in zip(
         axs, [img, circ, clahe_img, resized],
@@ -139,76 +113,22 @@ def preprocess_with_steps(img):
         ax.imshow(image)
         ax.set_title(title)
         ax.axis("off")
-
-    plt.tight_layout()
     st.pyplot(fig)
     plt.close(fig)
     return resized
 
-
 # --- Reasoning Text ---
 explanation_text = {
-    'Normal': """✅ **Normal**  <br>
-- 🟢 Clear retina, no lesions  <br>
-- 🩺 Blood vessels normal  <br>
-- 👁 Healthy optic disc & macula  <br>
-✔️ No signs of retinal disease.""",
-
-    'Diabetic Retinopathy': """💉 **Diabetic Retinopathy**  <br>
-- 🔶 Red spots / hemorrhages  <br>
-- 🩸 Leaking or swollen vessels  <br>
-- 👁 Macula possibly thickened  <br>
-⚠️ Indicative of diabetes-related damage.""",
-
-    'Glaucoma': """👁 **Glaucoma**  <br>
-- 🔴 Thinned nerve fiber layer  <br>
-- 💉 Cupping in optic disc  <br>
-- 👁 Risk of peripheral vision loss  <br>
-🔴 May need long-term eye pressure control.""",
-
-    'Cataract': """🌫 **Cataract**  <br>
-- 🌫 Cloudy or hazy image  <br>
-- 👁 Disc/macula not clearly visible  <br>
-- 🔍 Overall low contrast  <br>
-⚠️ Likely due to lens opacity.""",
-
-    'Age-related Macular Degeneration (AMD)': """🧓 **AMD**  <br>
-- 🔴 Yellow drusen near macula  <br>
-- 👁 Center vision affected  <br>
-- 🩺 Degenerative macula changes  <br>
-⚠️ Early to moderate AMD signs.""",
-
-    'Hypertension': """⚠️ **Hypertension**  <br>
-- 🔶 Bright lesions or hemorrhages  <br>
-- 🩸 Twisted/narrowed vessels  <br>
-- 👁 Star or flame-like patterns  <br>
-⚠️ Vascular damage from high BP.""",
-
-    'Myopia': """👓 **Myopia**  <br>
-- 🔵 Elongated eyeball signs  <br>
-- 🩺 Slight disc tilting  <br>
-- 👁 Possible peripapillary atrophy  <br>
-ℹ️ Common in nearsighted eyes.""",
-
-    'Others': """🔎 **Others**  <br>
-- ⚪ Unusual or unclassified patterns  <br>
-- 🩸 Irregular vascular changes  <br>
-- 👁 Disc or macula abnormalities  <br>
-❓ Possibly rare or overlapping conditions."""
+    'Normal': """✅ **Normal**<br>- 🟢 Clear retina, no lesions<br>- 🩺 Blood vessels normal<br>- 👁 Healthy optic disc & macula<br>✔️ No signs of retinal disease.""",
+    'Diabetic Retinopathy': """💉 **Diabetic Retinopathy**<br>- 🔶 Red spots / hemorrhages<br>- 🩸 Leaking or swollen vessels<br>- 👁 Macula possibly thickened<br>⚠️ Diabetes-related damage.""",
+    'Glaucoma': """👁 **Glaucoma**<br>- 🔴 Thinned nerve fiber layer<br>- 💉 Cupping in optic disc<br>- 👁 Risk of peripheral vision loss<br>🔴 Long-term eye pressure control needed.""",
+    'Cataract': """🌫 **Cataract**<br>- 🌫 Cloudy or hazy image<br>- 👁 Disc/macula not clearly visible<br>- 🔍 Overall low contrast<br>⚠️ Likely due to lens opacity.""",
+    'Age-related Macular Degeneration (AMD)': """🧓 **AMD**<br>- 🔴 Yellow drusen near macula<br>- 👁 Center vision affected<br>- 🩺 Degenerative macula changes<br>⚠️ Early/moderate AMD signs.""",
+    'Hypertension': """⚠️ **Hypertension**<br>- 🔶 Bright lesions or hemorrhages<br>- 🩸 Twisted/narrowed vessels<br>- 👁 Star or flame-like patterns<br>⚠️ Vascular damage from high BP.""",
+    'Myopia': """👓 **Myopia**<br>- 🔵 Elongated eyeball signs<br>- 🩺 Slight disc tilting<br>- 👁 Possible peripapillary atrophy<br>ℹ️ Common in nearsighted eyes.""",
+    'Others': """🔎 **Others**<br>- ⚪ Unusual or unclassified patterns<br>- 🩸 Irregular vascular changes<br>- 👁 Disc or macula abnormalities<br>❓ Possibly rare or overlapping conditions."""
 }
-# --- Prediction ---
-def predict(images, model):
-    images = np.array(images)
-    preds = model.predict(images, verbose=0)
-    if isinstance(preds, dict):  # Handle dict output (SavedModel case)
-        for v in preds.values():
-            if isinstance(v, (np.ndarray, list)):
-                return np.array(v)
-        return np.array(list(preds.values())[0])
-    else:
-        return preds
 
-# --- LIME Display ---
 # --- LIME Display ---
 def show_lime(img, model, pred_idx, pred_label, all_probs):
     with st.spinner("🟡 Generating LIME explanation..."):
@@ -224,44 +144,28 @@ def show_lime(img, model, pred_idx, pred_label, all_probs):
         )
         lime_img = mark_boundaries(temp, mask)
 
-        # Save to buffer once
         buf = BytesIO()
         plt.imsave(buf, lime_img, format="png")
         buf.seek(0)
         lime_data = buf.getvalue()
 
-        # Custom equal-height container
-        st.markdown('<div class="equal-cols">', unsafe_allow_html=True)
-
-        # Left column with image
-        with st.container():
+        col1, col2 = st.columns(2)
+        with col1:
             st.markdown("### 📍 LIME Explanation")
-            st.image(lime_data, use_container_width=True, output_format="PNG")
+            st.image(lime_data, width=224, output_format="PNG")  # 👈 Small LIME image
             st.download_button(
                 "📥 Download LIME Image",
                 lime_data,
                 file_name=f"{pred_label}_LIME.png",
                 mime="image/png"
             )
-
-        # Right column with explanation
-        with st.container():
+        with col2:
             st.markdown(
-                f"<div class='overlay-box'>"
-                f"<h3>🧠 Model's Reasoning</h3>"
-                f"{explanation_text.get(pred_label, 'No explanation available.')}"
-                "</div>",
+                f"<div class='overlay'>{explanation_text.get(pred_label, 'No explanation available.')}</div>",
                 unsafe_allow_html=True
             )
 
-        st.markdown('</div>', unsafe_allow_html=True)
-
-
-
-
-
-
-# --- Main App UI ---
+# --- Streamlit App UI ---
 st.set_page_config(page_title="👁 Retina Classifier with LIME", layout="wide")
 st.title("👁 Retina Disease Classifier with LIME Explanation")
 
@@ -294,4 +198,4 @@ if uploaded_files and selected_filename:
     st.success(f"✅ Prediction: **{pred_label}** ({confidence:.2f}%)")
     show_lime(preprocessed, model, pred_idx, pred_label, preds)
 else:
-    st.info("Upload retinal images from the sidebar to get started.")
+    st.info("📤 Upload a retinal image from the sidebar to get started.")