Update app.py

app.py

@@ -14,24 +14,24 @@ import datetime
 import zipfile
 from gradio.routes import Request
 
-# === SECRET ADMIN KEY ===
+# 🔐 Secret key
 ADMIN_KEY = "Diabetes_Detection"
 
-# === DEVICE SETUP ===
+# Device
 device = torch.device("cpu")
 
-# === MODEL LOADING ===
+# Load model
 model = models.resnet50(weights=None)
 model.fc = torch.nn.Linear(model.fc.in_features, 2)
 model.load_state_dict(torch.load("resnet50_dr_classifier.pth", map_location=device))
 model.to(device)
 model.eval()
 
-# === GRAD-CAM ===
+# Grad-CAM setup
 target_layer = model.layer4[-1]
 cam = GradCAM(model=model, target_layers=[target_layer])
 
-# === IMAGE TRANSFORM ===
+# Preprocessing
 transform = transforms.Compose([
     transforms.Resize((224, 224)),
     transforms.ToTensor(),
@@ -39,7 +39,7 @@ transform = transforms.Compose([
                          [0.229, 0.224, 0.225])
 ])
 
-# === STORAGE ===
+# Folders
 image_folder = "collected_images"
 os.makedirs(image_folder, exist_ok=True)
 
@@ -49,7 +49,7 @@ if not os.path.exists(csv_log_path):
         writer = csv.writer(f)
         writer.writerow(["timestamp", "image_filename", "prediction", "confidence"])
 
-# === PREDICT FUNCTION ===
+# 🔍 Prediction
 def predict_retinopathy(image):
     timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
     img = image.convert("RGB").resize((224, 224))
@@ -70,19 +70,18 @@ def predict_retinopathy(image):
     cam_image = show_cam_on_image(rgb_img_np, grayscale_cam, use_rgb=True)
     cam_pil = Image.fromarray(cam_image)
 
-    # Save uploaded image
+    # Save image + log
     image_filename = f"{timestamp}_{label.replace(' ', '_')}.png"
     image_path = os.path.join(image_folder, image_filename)
     image.save(image_path)
 
-    # Log prediction
     with open(csv_log_path, mode="a", newline="") as f:
         writer = csv.writer(f)
         writer.writerow([timestamp, image_filename, label, f"{confidence:.4f}"])
 
     return cam_pil, f"{label} (Confidence: {confidence:.2f})"
 
-# === ADMIN FILES ===
+# 📁 Admin downloads
 def download_csv():
     return csv_log_path
 
@@ -95,14 +94,13 @@ def download_dataset_zip():
             zipf.write(fpath, arcname=os.path.join("images", fname))
     return zip_filename
 
-# === VISIBILITY CHECK ===
+# ✅ Admin check (query param)
 def is_admin(request: Request):
-    query_params = dict(request.query_params)
-    return query_params.get("admin", "") == ADMIN_KEY
+    return request.query_params.get("admin") == ADMIN_KEY
 
-# === GRADIO APP ===
+# 🌐 App
 with gr.Blocks() as demo:
-    gr.Markdown("## 🧠 Diabetic Retinopathy Detection with Grad-CAM + Private Logging")
+    gr.Markdown("## 🧠 Diabetic Retinopathy Detection with Grad-CAM")
 
     with gr.Row():
         image_input = gr.Image(type="pil", label="Upload Retinal Image")
@@ -117,32 +115,26 @@ with gr.Blocks() as demo:
         outputs=[cam_output, prediction_output]
     )
 
+    # 🔒 Hidden admin section
     with gr.Column(visible=False) as admin_section:
-        gr.Markdown("### 🔐 Admin Downloads")
+        gr.Markdown("### 🔐 Private Downloads (Rodiyah Only)")
         with gr.Row():
             download_csv_btn = gr.Button("📄 Download CSV Log")
-            download_zip_btn = gr.Button("📦 Download Full Dataset")
+            download_zip_btn = gr.Button("📦 Download Dataset ZIP")
         csv_file = gr.File()
         zip_file = gr.File()
 
-    # Admin visibility only for Rodiyah with key
+    # ✅ Reveal only if correct ?admin=Diabetes_Detection in URL
     demo.load(
         lambda req: gr.update(visible=True) if is_admin(req) else gr.update(visible=False),
-        inputs=None,
+        inputs=[],
         outputs=admin_section,
-        queue=False
+        queue=False,
+        api_name=False,
     )
 
-    download_csv_btn.click(
-        fn=download_csv,
-        inputs=[],
-        outputs=csv_file
-    )
-
-    download_zip_btn.click(
-        fn=download_dataset_zip,
-        inputs=[],
-        outputs=zip_file
-    )
+    download_csv_btn.click(fn=download_csv, inputs=[], outputs=csv_file)
+    download_zip_btn.click(fn=download_dataset_zip, inputs=[], outputs=zip_file)
 
 demo.launch()
+