Upload folder using huggingface_hub

Dockerfile

@@ -1,3 +1,4 @@
+# Use a minimal base image with Python 3.9 installed
 FROM python:3.9
 
 # Set the working directory inside the container to /app
@@ -7,14 +8,12 @@ WORKDIR /app
 COPY requirements.txt .
 
 # Install Python dependencies listed in requirements.txt
-# Using --no-cache-dir to save space
 RUN pip install --no-cache-dir -r requirements.txt
 
-# --- CRITICAL FIX: Explicitly set Keras backend (optional but highly recommended) ---
-# This forces Keras to use the installed TensorFlow library.
+# --- CRITICAL FIX: Explicitly set Keras backend ---
 ENV KERAS_BACKEND=tensorflow
 
-# Copy application files and the model file
+# Copy application files and the model file (must be named tuned_ai_model_best_lat.keras)
 COPY app.py .
 COPY tuned_ai_model_best_lat.keras .
 
@@ -26,5 +25,6 @@ ENV HOME=/home/user \
 WORKDIR $HOME/app
 COPY --chown=user . $HOME/app
 
-# Define the command to run the Streamlit app
-CMD ["streamlit", "run", "app.py", "--server.port=7860", "--server.address=0.0.0.0"]
+# Define the command to run the Streamlit app, re-including the CRITICAL security flag
+# This flag should prevent the 403 Forbidden error caused by XSRF protection in proxy environments.
+CMD ["streamlit", "run", "app.py", "--server.port=7860", "--server.address=0.0.0.0", "--server.enableXsrfProtection=false"]

app.py

@@ -5,26 +5,25 @@ import tensorflow as tf
 from tensorflow.keras.models import load_model
 import os
 import time
-import requests
 from io import BytesIO
-import cv2 # Import cv2 for image resizing
 
 # Define the correct class names for output
 CLASS_NAMES = {0: 'Normal', 1: 'Viral Pneumonia', 2: 'Covid'}
-IMG_SIZE = 224 # Set this constant correctly
+IMG_SIZE = 224 
 
 # Function to load the model (cached for efficiency)
 @st.cache_resource
 def load_tuned_model():
-    # Attempt loading the model
-    # Use the correct path to the saved model
+    # Use custom_objects for maximum robustness when loading models with 
+    # pre-trained components like VGG16 in specific environments.
     return tf.keras.models.load_model(
         "tuned_ai_model_best_lat.keras",
-        # custom_objects are not needed for this model
+        custom_objects={'VGG16': tf.keras.applications.VGG16}
     )
 
 # Function to run the prediction and show progress
 def run_prediction(image_file, model, img_size):
+    # Progress visualization (for user experience)
     progress_bar = st.progress(0)
     status_text = st.empty()
     for i in range(100):
@@ -33,34 +32,46 @@ def run_prediction(image_file, model, img_size):
         time.sleep(0.01)
 
     try:
-        # Read the image file
+        # Read and process the image using PIL (Image)
         image = Image.open(image_file).convert("RGB")
-
-        # Correct image resizing to match model's trained input shape (224x224)
         img_array = np.array(image.resize((img_size, img_size)))
-
-        # Expand dimensions to create a batch of size 1
         img_array = np.expand_dims(img_array, axis=0)
-
-        # Normalize the image data
         img_array = img_array / 255.0
 
         # Make the prediction
         prediction = model.predict(img_array).flatten()
 
-        # Correctly find the index of the class with the highest probability (0, 1, or 2)
+        # Find the predicted class index and name
         class_predicted_idx = np.argmax(prediction)
         predicted_class_name = CLASS_NAMES[class_predicted_idx]
 
         status_text.success("Prediction complete!")
-        return predicted_class_name, prediction
-
+        
+        # Display results
+        st.subheader("Prediction Probabilities:")
+        for i, prob in enumerate(prediction):
+            class_name = CLASS_NAMES[i]
+            if i == class_predicted_idx:
+                st.markdown(f"**{class_name}: {prob*100:.2f}%** (Predicted)", unsafe_allow_html=True)
+            else:
+                st.markdown(f"{class_name}: {prob*100:.2f}%")
+
+        if class_predicted_idx == 2:
+            st.error(f"**Result: ❗ HIGH LIKELIHOOD OF COVID DETECTED ❗**")
+        else:
+            st.success(f"**Result: ✅ {predicted_class_name} Detected**")
+        
     except Exception as e:
         status_text.error(f"An error occurred during prediction: {e}")
-        return None, None
+        st.exception(e) # Show full traceback for debugging
 
-# Load the model
-model = load_tuned_model()
+# Load the model and ensure the app stops if loading fails
+try:
+    model = load_tuned_model()
+except Exception as e:
+    st.error("Model Loading Failed. Check TF/Keras versions and model file name (tuned_ai_model_best_lat.keras).")
+    st.exception(e)
+    st.stop()
 
 st.title("COVID Detection from Chest X-ray")
 st.write("Upload a chest X-ray image to predict if it shows signs of Normal, Viral Pneumonia, or COVID.")
@@ -68,17 +79,8 @@ st.write("Upload a chest X-ray image to predict if it shows signs of Normal, Vir
 uploaded_file = st.file_uploader("Choose an X-ray image...", type=["jpg", "jpeg", "png"])
 
 if uploaded_file is not None:
-    # Display the uploaded image
     image = Image.open(uploaded_file)
     st.image(image, caption="Uploaded X-ray Image", use_container_width=True)
 
-    # Make prediction when button is clicked
     if st.button("Predict"):
-        predicted_class, prediction_probabilities = run_prediction(uploaded_file, model, IMG_SIZE)
-
-        if predicted_class:
-            st.write(f"Prediction: **{predicted_class}**")
-            st.write("Prediction Probabilities:")
-            # Display probabilities for all classes
-            for i, prob in enumerate(prediction_probabilities):
-                st.write(f"- {CLASS_NAMES[i]}: {prob:.4f}")
+        run_prediction(uploaded_file, model, IMG_SIZE)

requirements.txt

@@ -4,4 +4,3 @@ numpy==2.0.2
 pandas==2.2.2
 keras==3.10.0
 Pillow==11.3.0
-opencv-python==4.12.0.88