Update app.py

app.py

@@ -27,30 +27,6 @@ IMG_SIZE = (224, 224)
 CLASS_NAMES = ['Normal', 'Diabetes', 'Glaucoma', 'Cataract', 'AMD', 'Hypertension', 'Myopia', 'Others']
 LIME_EXPLAINER = lime_image.LimeImageExplainer()
 
-# --- Enhanced LIME Explanation Dictionary (HTML + Icons + Colors) ---
-explanation_text = {
-    'Normal': """✅ **<span style='color:green'>Normal</span>**\nThe model predicted <span style='color:green'><strong>Normal</strong></span> because everything looks healthy:\n- 🟢 **Overall Look**: Clear retina, no spots or damage  \n- 🩺 **Blood Vessels**: Normal size and shape  \n- 👁 **Optic Disc & Macula**: Clean, no swelling or dark areas  \n<span style='color:green'>✔️ The eye appears completely healthy.</span>""",
-    'Diabetes': """💉 **<span style='color:orange'>Diabetes</span>**\nThe model predicted <span style='color:orange'><strong>Diabetes</strong></span> due to signs of diabetic eye disease:\n- 🔶 **Overall Look**: Tiny spots, bleeding, or swelling  \n- 🩸 **Blood Vessels**: Damaged or swollen  \n- 👁 **Optic Disc & Macula**: Swelling or fluid may be present  \n<span style='color:orange'>⚠️ Signs of diabetic retinopathy.</span>""",
-    'Glaucoma': """👁 **<span style='color:red'>Glaucoma</span>**\nThe model predicted <span style='color:red'><strong>Glaucoma</strong></span> by detecting optic nerve changes:\n- 🔴 **Overall Look**: Some nerve areas look damaged  \n- 💉 **Blood Vessels**: Mostly normal  \n- 👁 **Optic Disc & Macula**: Disc cupping, nerve fiber loss  \n<span style='color:red'>🔴 Serious condition—can lead to vision loss.</span>""",
-    'Cataract': """🌫 **<span style='color:gray'>Cataract</span>**\nThe model predicted <span style='color:gray'><strong>Cataract</strong></span> based on blurry image quality:\n- 🌫 **Overall Look**: Image appears cloudy or foggy  \n- 👁 **Blood Vessels**: Hard to see clearly  \n- 🔍 **Optic Disc & Macula**: Low contrast and unclear  \n<span style='color:gray'>⚠️ Likely due to a cloudy lens (cataract).</span>""",
-    'AMD': """🧓 **<span style='color:red'>AMD (Macular Degeneration)</span>**\nThe model predicted <span style='color:red'><strong>AMD</strong></span> by focusing on the center of the retina:\n- 🔴 **Overall Look**: Mostly clear except center  \n- 🩺 **Blood Vessels**: Usually look fine  \n- 👁 **Optic Disc & Macula**: Yellow spots or damage in macula  \n<span style='color:red'>⚠️ Early signs of age-related vision loss.</span>""",
-    'Hypertension': """⚠️ **<span style='color:orange'>Hypertension</span>**\nThe model predicted <span style='color:orange'><strong>Hypertension</strong></span> due to blood vessel changes:\n- 🔶 **Overall Look**: Small bleeds or spots  \n- 🩸 **Blood Vessels**: Narrowed or twisted  \n- 👁 **Optic Disc & Macula**: Swelling or star patterns  \n<span style='color:orange'>⚠️ May be caused by high blood pressure.</span>""",
-    'Myopia': """👓 **<span style='color:blue'>Myopia</span>**\nThe model predicted <span style='color:blue'><strong>Myopia</strong></span> based on eye structure:\n- 🔵 **Overall Look**: Stretched or thin retina edges  \n- 🩺 **Blood Vessels**: Usually normal  \n- 👁 **Optic Disc & Macula**: Tilted disc, outer damage  \n<span style='color:blue'>ℹ️ Common in people with nearsightedness.</span>""",
-    'Others': """🔎 **<span style='color:gray'>Others</span>**\nThe model predicted <span style='color:gray'><strong>Others</strong></span> due to unknown or rare changes:\n- ⚪ **Overall Look**: Mixed or unusual patterns  \n- 🩸 **Blood Vessels**: Varies by case  \n- 👁 **Optic Disc & Macula**: Random changes  \n<span style='color:gray'>❓ May indicate a rare or mixed eye condition.</span>"""
-}
-
-# --- Streamlit Setup ---
-st.set_page_config(page_title="👁 Retina Classifier - Multi Image LIME", layout="wide")
-st.title("👁 Retina Disease Classifier with LIME Explanation")
-st.markdown("""
-    <style>
-    .lime-text {
-        color: var(--text-color);
-        font-size: 16px;
-    }
-    </style>
-""", unsafe_allow_html=True)
-
 @st.cache_resource
 def load_model():
     model_path = "Model"
@@ -81,8 +57,11 @@ def preprocess_with_steps(img):
     resized = cv2.resize(sharp, IMG_SIZE) / 255.0
 
     fig, axs = plt.subplots(1, 4, figsize=(20, 5))
-    for ax, image, title in zip(axs, [img, circ, clahe_img, resized],
-                                 ["Original", "Circular Crop", "CLAHE", "Sharpen + Resize"]):
+    for ax, image, title in zip(
+        axs,
+        [img, circ, clahe_img, resized],
+        ["Original", "Circular Crop", "CLAHE", "Sharpen + Resize"],
+    ):
         ax.imshow(image)
         ax.set_title(title)
         ax.axis("off")
@@ -95,6 +74,65 @@ def predict(images, model):
     preds = model.predict(images, verbose=0)
     return list(preds.values())[0] if isinstance(preds, dict) else preds
 
+# --- Easy Language LIME Explanation Dictionary ---
+explanation_text = {
+    'Normal': """✅ **<span style='color:green'>Normal</span>**<br>
+The model predicted <strong>Normal</strong> because everything looks healthy:<br>
+- 🟢 Overall retina looks clear, no spots or damage<br>
+- 🩺 Blood vessels have normal size and shape<br>
+- 👁 Optic disc and macula are clean and well-defined<br>
+<span style='color:green'>✔️ This is a healthy eye.</span>""",
+
+    'Diabetes': """💉 **<span style='color:orange'>Diabetes</span>**<br>
+The model detected signs of diabetic eye disease:<br>
+- 🔶 Tiny red spots, bleeding, or swelling<br>
+- 🩸 Blood vessels appear damaged or thickened<br>
+- 👁 Macula may show fluid or swelling<br>
+<span style='color:orange'>⚠️ Diabetic retinopathy detected.</span>""",
+
+    'Glaucoma': """👁 **<span style='color:red'>Glaucoma</span>**<br>
+The model noticed optic nerve damage patterns:<br>
+- 🔴 Some nerve areas seem thin or missing<br>
+- 💉 Blood vessels are mostly normal<br>
+- 👁 Disc cupping and nerve fiber loss<br>
+<span style='color:red'>🔴 May cause permanent vision loss if untreated.</span>""",
+
+    'Cataract': """🌫 **<span style='color:gray'>Cataract</span>**<br>
+The image quality was affected by lens clouding:<br>
+- 🌫 Whole image looks hazy or foggy<br>
+- 👁 Blood vessels are not clearly visible<br>
+- 🔍 Macula and disc are blurry<br>
+<span style='color:gray'>⚠️ Cataract is likely reducing image clarity.</span>""",
+
+    'AMD': """🧓 **<span style='color:red'>AMD (Macular Degeneration)</span>**<br>
+The model found damage in the central retina (macula):<br>
+- 🔴 Center of retina shows yellow spots or damage<br>
+- 🩺 Blood vessels usually look fine<br>
+- 👁 Macula shows signs of age-related change<br>
+<span style='color:red'>⚠️ Early signs of macular degeneration.</span>""",
+
+    'Hypertension': """⚠️ **<span style='color:orange'>Hypertension</span>**<br>
+The model found signs of high blood pressure in the retina:<br>
+- 🔶 Small bleeds or bright spots in retina<br>
+- 🩸 Blood vessels narrowed or twisted<br>
+- 👁 Possible swelling or star patterns<br>
+<span style='color:orange'>⚠️ Hypertensive retinopathy suspected.</span>""",
+
+    'Myopia': """👓 **<span style='color:blue'>Myopia</span>**<br>
+The model identified signs of nearsightedness:<br>
+- 🔵 Retinal edges look stretched or thin<br>
+- 🩺 Blood vessels are mostly normal<br>
+- 👁 Tilted optic disc, possible outer damage<br>
+<span style='color:blue'>ℹ️ Common in high myopia cases.</span>""",
+
+    'Others': """🔎 **<span style='color:gray'>Others</span>**<br>
+Mixed or unclear patterns seen in retina:<br>
+- ⚪ Random spots or patches<br>
+- 🩸 Blood vessel features vary<br>
+- 👁 Disc or macula changes are not typical<br>
+<span style='color:gray'>❓ Could be a rare or unclassified condition.</span>"""
+}
+
 def show_lime(img, model, pred_idx, pred_label):
     with st.spinner("🟡 LIME explanation is loading..."):
         explanation = LIME_EXPLAINER.explain_instance(
@@ -108,25 +146,28 @@ def show_lime(img, model, pred_idx, pred_label):
             label=pred_idx, positive_only=True, num_features=10, hide_rest=False
         )
 
-        fig, ax = plt.subplots(figsize=(3, 3))
-        ax.imshow(mark_boundaries(temp, mask))
-        ax.axis("off")
-        ax.text(5, 20, f"LIME: {pred_label}", color="black", fontsize=10,
-                bbox=dict(facecolor='white', alpha=0.7, pad=2))
-        st.pyplot(fig)
-
-        buf = BytesIO()
-        fig.savefig(buf, format="png")
-        st.download_button("📥 Download LIME Image", buf.getvalue(), file_name=f"{pred_label}_LIME.png", mime="image/png")
-        plt.close(fig)
-
-        st.markdown("#### 👁 Detailed LIME Explanation")
-        st.markdown(
-            f"<div class='lime-text'>{explanation_text.get(pred_label, 'No explanation available.')}</div>",
-            unsafe_allow_html=True
-        )
+        col1, col2 = st.columns([1, 2])
+
+        with col1:
+            fig, ax = plt.subplots(figsize=(4, 4))
+            ax.imshow(mark_boundaries(temp, mask))
+            ax.axis("off")
+            ax.set_title("📍 LIME Highlighted Areas", fontsize=12)
+            st.pyplot(fig)
+
+            buf = BytesIO()
+            fig.savefig(buf, format="png")
+            st.download_button("📥 Download LIME Image", buf.getvalue(), file_name=f"{pred_label}_LIME.png", mime="image/png")
+            plt.close(fig)
+
+        with col2:
+            st.markdown("### 🧠 Model's Reasoning")
+            st.markdown(explanation_text.get(pred_label, "No explanation available."), unsafe_allow_html=True)
+
+# --- UI ---
+st.set_page_config(page_title="👁 Retina Classifier - Multi Image LIME", layout="wide")
+st.title("👁 Retina Disease Classifier with LIME Explanation")
 
-# --- Sidebar + File Upload ---
 model = load_model()
 
 with st.sidebar:
@@ -152,23 +193,4 @@ if uploaded_files and selected_filename:
     confidence = np.max(preds) * 100
 
     st.success(f"✅ Prediction: **{pred_label}** ({confidence:.2f}%)")
-
-    # Show probability chart
-    st.markdown("#### 📊 Class Probabilities")
-    prob_dict = {label: float(preds[0][i]) for i, label in enumerate(CLASS_NAMES)}
-    prob_chart_data = sorted(prob_dict.items(), key=lambda x: x[1], reverse=True)
-
-    prob_labels = [k for k, _ in prob_chart_data]
-    prob_values = [v for _, v in prob_chart_data]
-
-    fig_bar, ax_bar = plt.subplots(figsize=(8, 3))
-    bars = ax_bar.barh(prob_labels, prob_values, color="skyblue")
-    ax_bar.set_xlim(0, 1)
-    ax_bar.invert_yaxis()
-    ax_bar.set_xlabel("Confidence Score")
-    for bar in bars:
-        ax_bar.text(bar.get_width() + 0.01, bar.get_y() + 0.25, f"{bar.get_width()*100:.1f}%", color='black')
-    st.pyplot(fig_bar)
-    plt.close(fig_bar)
-
     show_lime(preprocessed, model, pred_idx, pred_label)