Update app.py

app.py

@@ -1,5 +1,5 @@
 import os
-# Force Keras 2 logic to prevent the 'recursion depth' and 'quantization' errors
+# Forces the use of Keras 2 logic, preventing common 'recursion' errors in new environments
 os.environ["TF_USE_LEGACY_KERAS"] = "1" 
 
 import gradio as gr
@@ -10,12 +10,11 @@ from PIL import Image
 from huggingface_hub import hf_hub_download
 
 # --- CONFIGURATION ---
-# 1. Ensure this matches your Hugging Face URL exactly. 
-# If your repo is at huggingface.co/username/repo, use "username/repo"
-REPO_ID = "mediaportal/BrainTumorMRI" 
+# Based on your provided link: https://huggingface.co/mediaportal/Braintumor-MRI-detection
+REPO_ID = "mediaportal/Braintumor-MRI-detection" 
 MODEL_FILENAME = "BraintumorMRI99.h5"
 
-# 2. If your repo is PRIVATE, add your token to Space Secrets as 'HF_TOKEN'
+# If your repo is set to PRIVATE, add your token to Space Secrets as 'HF_TOKEN'
 hf_token = os.getenv("HF_TOKEN")
 
 model = None
@@ -23,17 +22,15 @@ model = None
 def load_model_with_progress(progress=gr.Progress(track_tqdm=True)):
     global model
     try:
-        progress(0, desc="Connecting to Hugging Face...")
-        
-        # hf_hub_download will use the token if the repo is private
+        progress(0, desc="Downloading model from Hugging Face...")
         path = hf_hub_download(
             repo_id=REPO_ID, 
             filename=MODEL_FILENAME,
             token=hf_token
         )
         
-        progress(0.7, desc="Loading Xception weights (Keras 2)...")
-        # compile=False is used to skip loading training-specific optimizer states
+        progress(0.7, desc="Loading weights into Xception architecture...")
+        # compile=False avoids loading the optimizer state from the training session
         model = keras.models.load_model(path, compile=False)
         
         progress(1.0, desc="✅ Model Ready!")
@@ -43,48 +40,42 @@ def load_model_with_progress(progress=gr.Progress(track_tqdm=True)):
 
 def predict(img):
     if model is None:
-        return "Please wait for the model to finish loading."
+        return "System is still initializing. Please wait."
     
     if img is None:
         return "No image provided."
 
-    # 1. Preprocessing: Xception requires 299x299 based on your notebook
+    # Preprocessing based on notebook: Xception requires 299x299
     img = Image.fromarray(img.astype('uint8'), 'RGB').resize((299, 299))
     
-    # 2. Rescale 1/255 as used in your ImageDataGenerator
+    # Rescale 1/255 as used in the notebook's ImageDataGenerator
     img_array = np.array(img).astype('float32') / 255.0
-    
-    # 3. Add batch dimension (1, 299, 299, 3)
     img_array = np.expand_dims(img_array, axis=0)
     
-    # Inference
     prediction = model.predict(img_array)[0]
     
-    # Labels based on the 4 classes in your training data
+    # Class labels identified from tr_df in the notebook
+    # Standard order for this dataset: glioma, meningioma, notumor, pituitary
     labels = ["Glioma", "Meningioma", "No Tumor", "Pituitary"]
     
     return {labels[i]: float(prediction[i]) for i in range(len(labels))}
 
 # --- GRADIO INTERFACE ---
-
-
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
-    gr.Markdown("# 🧠 Brain Tumor MRI Identification")
-    gr.Markdown("Identify Glioma, Meningioma, Pituitary tumors or Healthy scans using Xception Deep Learning.")
+    gr.Markdown("# 🧠 Brain Tumor MRI Classification")
+    gr.Markdown("Identify Glioma, Meningioma, Pituitary tumors, or Healthy scans.")
     
-    status_box = gr.Markdown("⏳ Initializing system...")
+    status_box = gr.Markdown("⏳ Initializing... Checking access to " + REPO_ID)
     
     with gr.Row():
         with gr.Column():
             input_img = gr.Image(label="Upload MRI Scan")
-            btn = gr.Button("Analyze MRI", variant="primary")
+            btn = gr.Button("Run Diagnosis", variant="primary")
         with gr.Column():
-            output_label = gr.Label(num_top_classes=4, label="Diagnosis Prediction")
+            output_label = gr.Label(num_top_classes=4, label="Prediction Result")
 
-    # Load model on start
+    # Load model immediately upon app startup
     demo.load(load_model_with_progress, outputs=status_box)
-    
-    # Run prediction on click
     btn.click(fn=predict, inputs=input_img, outputs=output_label)
 
 if __name__ == "__main__":