Update app.py

app.py

@@ -1,140 +1,168 @@
+import os
+import traceback
+import numpy as np
 import streamlit as st
-from transformers import pipeline
 from PIL import Image
-import requests
+from transformers import pipeline
+import matplotlib.pyplot as plt
+from skimage.color import rgb2gray
+from skimage.filters import threshold_otsu
+
+
+# =======================
+# Configuration and Setup
+# =======================
+
+# Streamlit Page Configuration
+st.set_page_config(
+    page_title="AI Cancer Detection Platform",
+    page_icon="🩺",
+    layout="wide",
+    initial_sidebar_state="expanded",
+    menu_items={
+        "About": "### AI Cancer Detection Platform\n"
+                 "Developed to classify cancer images and provide research insights."
+    }
+)
+
 
 # =======================
-# Caching the Model
+# Helper Functions
 # =======================
+
 @st.cache_resource
-def load_model():
+def load_pipeline():
     """
-    Load the pre-trained skin cancer classification model.
-    Cached to prevent reloading on every app interaction.
+    Load the pre-trained image classification pipeline using PyTorch as the backend.
     """
-    return pipeline("image-classification", model="Anwarkh1/Skin_Cancer-Image_Classification")
+    try:
+        model_pipeline = pipeline(
+            "image-classification",
+            model="Anwarkh1/Skin_Cancer-Image_Classification",
+            framework="pt"  # Force PyTorch backend
+        )
+        return model_pipeline
+    except Exception as e:
+        st.error(f"Error loading model: {e}")
+        traceback.print_exc()
+        st.stop()
 
-model = load_model()
 
-# =======================
-# Functionality: Classify Skin Cancer
-# =======================
-def classify_skin_cancer(image):
-    results = model(image)
-    label = results[0]['label']
-    confidence = results[0]['score']
-    explanation = f"The model predicts **{label}** with a confidence of {confidence:.2%}."
-    return label, confidence, explanation
+def process_image(image):
+    """
+    Perform image processing to extract features for better visualization.
+    """
+    try:
+        # Convert image to grayscale
+        gray_image = rgb2gray(np.array(image))
 
-# =======================
-# Functionality: Fetch Cancer Research Papers
-# =======================
-@st.cache_data
-def fetch_cancer_research():
+        # Apply Otsu's threshold
+        thresh = threshold_otsu(gray_image)
+        binary = gray_image > thresh
+
+        # Calculate edge pixel percentage
+        edge_pixels = np.sum(binary)
+        total_pixels = binary.size
+        edge_percentage = (edge_pixels / total_pixels) * 100
+
+        # Generate plots
+        fig, ax = plt.subplots(1, 2, figsize=(10, 5))
+        ax[0].imshow(gray_image, cmap="gray")
+        ax[0].set_title("Grayscale Image")
+        ax[0].axis("off")
+
+        ax[1].imshow(binary, cmap="gray")
+        ax[1].set_title("Binary Image (Thresholded)")
+        ax[1].axis("off")
+
+        plt.tight_layout()
+        st.pyplot(fig)
+
+        # Feature description
+        return f"{edge_percentage:.2f}% of the image contains edge pixels after thresholding."
+
+    except Exception as e:
+        st.error(f"Error processing image: {e}")
+        traceback.print_exc()
+        return "No significant features extracted."
+
+
+def classify_image(image, model_pipeline):
     """
-    Fetch the latest research papers related to skin cancer.
-    Cached to avoid repeated API calls.
+    Classify the uploaded image using the pre-trained model pipeline.
     """
-    api_url = "https://api.semanticscholar.org/graph/v1/paper/search"
-    params = {
-        "query": "skin cancer research",
-        "fields": "title,abstract,url",
-        "limit": 5
-    }
-    response = requests.get(api_url, params=params)
-    if response.status_code == 200:
-        papers = response.json().get("data", [])
-        summaries = []
-        for paper in papers:
-            title = paper.get("title", "No Title")
-            abstract = paper.get("abstract", "No Abstract")
-            url = paper.get("url", "No URL")
-            summaries.append(f"**{title}**\n\n{abstract}\n\n[Read More]({url})")
-        return "\n\n---\n\n".join(summaries)
-    else:
-        return "Error fetching research papers. Please try again later."
+    try:
+        # Resize image to 224x224 as required by the model
+        image_resized = image.resize((224, 224))
+        predictions = model_pipeline(image_resized)
 
-# =======================
-# Streamlit Page Config
-# =======================
-st.set_page_config(
-    page_title="AI-Powered Skin Cancer Detection",
-    page_icon="🩺",
-    layout="wide",
-    initial_sidebar_state="expanded"
-)
+        if predictions:
+            top_prediction = predictions[0]
+            label = top_prediction["label"]
+            score = top_prediction["score"]
+            return label, score
+        else:
+            st.warning("No predictions were made.")
+            return None, None
+    except Exception as e:
+        st.error(f"Error during classification: {e}")
+        traceback.print_exc()
+        return None, None
 
-st.sidebar.header("Navigation")
-app_mode = st.sidebar.radio(
-    "Choose a feature",
-    ["🔍 Skin Cancer Classification", "📄 Latest Research Papers", "ℹ️ About the Model"]
-)
 
 # =======================
-# Skin Cancer Classification
+# Streamlit Main Content
 # =======================
-if app_mode == "🔍 Skin Cancer Classification":
-    st.title("🔍 Skin Cancer Classification")
-    st.write(
-        "Upload an image of the skin lesion, and the AI model will classify it as one of several types, "
-        "such as melanoma, basal cell carcinoma, or benign keratosis-like lesions."
-    )
 
-    uploaded_image = st.file_uploader("Upload a skin lesion image", type=["png", "jpg", "jpeg"])
+st.title("🩺 AI-Powered Cancer Detection")
 
-    if uploaded_image:
-        image = Image.open(uploaded_image).convert('RGB')
+# Image Upload Section
+st.subheader("📤 Upload a Cancer Image")
+uploaded_image = st.file_uploader("Choose an image file...", type=["png", "jpg", "jpeg"])
+
+if uploaded_image is not None:
+    try:
+        # Open the uploaded image
+        image = Image.open(uploaded_image).convert("RGB")
+
+        # Display the uploaded image
         st.image(image, caption="Uploaded Image", use_column_width=True)
 
-        # Perform classification
-        st.write("Classifying...")
-        label, confidence, explanation = classify_skin_cancer(image)
+        # Process the image
+        st.markdown("### 🛠️ Image Processing")
+        processed_features = process_image(image)
 
-        # Display results
-        st.markdown(f"### **Prediction**: {label}")
-        st.markdown(f"### **Confidence**: {confidence:.2%}")
-        st.markdown(f"### **Explanation**: {explanation}")
+        # Load the model pipeline
+        st.markdown("### 🔍 Classifying the Image")
+        model_pipeline = load_pipeline()
 
-# =======================
-# Latest Research Papers
-# =======================
-elif app_mode == "📄 Latest Research Papers":
-    st.title("📄 Latest Research Papers")
-    st.write(
-        "Fetch the latest research papers on skin cancer to stay updated on recent findings and innovations."
-    )
+        # Classify the image
+        with st.spinner("Classifying..."):
+            label, confidence = classify_image(image, model_pipeline)
 
-    if st.button("Fetch Papers"):
-        with st.spinner("Fetching research papers..."):
-            summaries = fetch_cancer_research()
-        st.markdown(summaries)
+        if label and confidence:
+            st.write(f"**Prediction:** {label}")
+            st.write(f"**Confidence:** {confidence:.2%}")
 
-# =======================
-# About the Model
-# =======================
-elif app_mode == "ℹ️ About the Model":
-    st.title("ℹ️ About the Skin Cancer Detection Model")
-    st.markdown("""
-    - **Model Architecture:** Vision Transformer (ViT)  
-    - **Trained On:** Skin Cancer Dataset (ISIC)  
-    - **Classes:**  
-      - Benign keratosis-like lesions  
-      - Basal cell carcinoma  
-      - Actinic keratoses  
-      - Vascular lesions  
-      - Melanocytic nevi  
-      - Melanoma  
-      - Dermatofibroma  
-    - **Performance Metrics:**  
-      - **Validation Accuracy:** 96.95%  
-      - **Train Accuracy:** 96.14%  
-    - **Loss Function:** Cross-Entropy  
-    """)
+            # Highlight prediction confidence
+            if confidence > 0.80:
+                st.success("High confidence in the prediction.")
+            elif confidence > 0.50:
+                st.warning("Moderate confidence in the prediction.")
+            else:
+                st.error("Low confidence in the prediction.")
+
+    except Exception as e:
+        st.error(f"An unexpected error occurred: {e}")
+        traceback.print_exc()
+else:
+    st.info("Upload an image to start the classification.")
 
 # =======================
 # Footer
 # =======================
-st.sidebar.info("""
-Developed by **[mgbam](https://huggingface.co/mgbam)**  
-This app leverages state-of-the-art AI models for skin cancer detection and research insights.
+
+st.markdown("""
+---
+**AI Cancer Detection Platform** | This application is for informational purposes only and is not intended for medical diagnosis.
 """)