Upload folder using huggingface_hub

app.py

@@ -33,7 +33,7 @@ class CometDetectorV3:
             
             fits_files = sorted(Path(tmpdir).rglob('*.fts')) + sorted(Path(tmpdir).rglob('*.fits'))
             
-            for fpath in fits_files:
+            for fpath in fits_files[:50]:  # Limit to 50 files to prevent timeout
                 try:
                     with fits.open(fpath) as hdul:
                         img = hdul[0].data.astype(np.float32)
@@ -56,20 +56,12 @@ class CometDetectorV3:
         return max_proj
     
     def classify_image(self, max_proj):
-        # Normalize
         img = (max_proj - max_proj.min()) / (max_proj.max() - max_proj.min() + 1e-8)
-        
-        # Resize to 512x512
         img = cv2.resize(img, (512, 512))
-        
-        # To RGB
         img_rgb = np.stack([img, img, img], axis=0)
         img_tensor = torch.FloatTensor(img_rgb).unsqueeze(0).to(self.device)
-        
-        # Normalize
         img_tensor = self.transform(img_tensor)
         
-        # Predict
         with torch.no_grad():
             output = self.model(img_tensor)
             probs = torch.softmax(output, dim=1)
@@ -80,100 +72,235 @@ class CometDetectorV3:
     
     def process_and_visualize(self, zip_file):
         try:
+            if zip_file is None:
+                return None, "⚠️ Please upload a ZIP file"
+            
             images = self.load_fits_from_zip(zip_file.name)
             
             if len(images) < 2:
-                return None, "❌ Need at least 2 FITS images"
+                return None, "❌ Need at least 2 FITS images in the ZIP file"
             
             max_proj = self.create_difference_images(images)
             pred_class, confidence = self.classify_image(max_proj)
             
-            # Visualize
-            fig, axes = plt.subplots(1, 2, figsize=(14, 6))
+            # Create figure with dark background
+            plt.style.use('dark_background')
+            fig, axes = plt.subplots(1, 2, figsize=(16, 7))
+            fig.patch.set_facecolor('#0B1120')
             
+            # First image
             axes[0].imshow(images[0], cmap='gray')
-            axes[0].set_title('First Image in Sequence')
+            axes[0].set_title('Original SOHO/LASCO C3 Image', 
+                            fontsize=13, color='#E0E0E0', pad=15)
             axes[0].axis('off')
             
+            # Difference projection
             axes[1].imshow(max_proj, cmap='hot')
             if pred_class == 1:
-                axes[1].set_title('⭐ COMET DETECTED', fontsize=14, color='green', weight='bold')
+                axes[1].set_title('🌟 COMET SIGNATURE DETECTED', 
+                                fontsize=14, color='#00FF88', weight='bold', pad=15)
             else:
-                axes[1].set_title('Background - No Comet', fontsize=14)
+                axes[1].set_title('Background Sequence', 
+                                fontsize=13, color='#FFB366', pad=15)
             axes[1].axis('off')
             
             plt.tight_layout()
             
-            # Summary
+            # Summary with enhanced styling
             if pred_class == 1:
-                summary = f"## ✅ **COMET DETECTED!**\n\n"
-                summary += f"**Confidence:** {confidence:.1%}\n\n"
-                summary += "This sequence shows signs of comet activity.\n\n"
+                summary = f"""
+# 🌟 COMET DETECTED!
+
+### Detection Confidence: {confidence:.1%}
+
+---
+
+**Analysis Results:**
+- **Classification:** Comet Event ✅
+- **Confidence Score:** {confidence:.3f}
+- **Images Analyzed:** {len(images)}
+- **Detection Method:** Difference Imaging + CNN
+
+**Interpretation:**  
+This sequence shows characteristic signatures of a sungrazing comet passing through 
+the SOHO/LASCO C3 field of view. The bright trail in the difference projection 
+indicates significant motion and brightness changes consistent with comet activity.
+
+---
+*Model: EfficientNet-B0 | Accuracy: 97.7%*
+"""
             else:
-                summary = f"## ❌ **No Comet Detected**\n\n"
-                summary += f"**Confidence:** {confidence:.1%}\n\n"
-                summary += "This appears to be a background sequence.\n\n"
-            
-            summary += f"---\n**Processing Info:**\n"
-            summary += f"- Images analyzed: {len(images)}\n"
-            summary += f"- Model: EfficientNet-B0\n"
-            summary += f"- Accuracy: 97.7%"
+                summary = f"""
+# 🌑 No Comet Detected
+
+### Background Confidence: {confidence:.1%}
+
+---
+
+**Analysis Results:**
+- **Classification:** Background Sequence ⚪
+- **Confidence Score:** {confidence:.3f}
+- **Images Analyzed:** {len(images)}
+- **Detection Method:** Difference Imaging + CNN
+
+**Interpretation:**  
+This sequence does not show signatures consistent with comet activity. 
+The difference projection reveals minimal motion or brightness changes, 
+typical of background coronagraph observations.
+
+---
+*Model: EfficientNet-B0 | Accuracy: 97.7%*
+"""
             
             return fig, summary
             
         except Exception as e:
-            return None, f"❌ Error: {str(e)}"
+            return None, f"❌ **Processing Error**\n\n{str(e)}\n\nPlease ensure your ZIP contains valid SOHO FITS files."
 
 detector = CometDetectorV3('best_model.pth')
 
-with gr.Blocks(theme=gr.themes.Soft(), title="SOHO Comet Detector") as demo:
-    gr.Markdown("""
-    # 🌟 AI-Powered SOHO Comet Detector
+# Custom CSS for astronomy theme
+custom_css = """
+@import url('https://fonts.googleapis.com/css2?family=Space+Grotesk:wght@400;600;700&display=swap');
+
+body {
+    font-family: 'Space Grotesk', sans-serif !important;
+}
+
+.gradio-container {
+    background: linear-gradient(135deg, #0B1120 0%, #1a1f35 100%) !important;
+}
+
+.contain {
+    background: rgba(15, 20, 35, 0.8) !important;
+    backdrop-filter: blur(10px) !important;
+    border: 1px solid rgba(255, 255, 255, 0.1) !important;
+    border-radius: 16px !important;
+}
+
+h1 {
+    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+    -webkit-background-clip: text;
+    -webkit-text-fill-color: transparent;
+    font-weight: 700 !important;
+    font-size: 3em !important;
+    text-align: center !important;
+    margin-bottom: 1em !important;
+}
+
+.gr-button-primary {
+    background: linear-gradient(135deg, #667eea 0%, #764ba2 100%) !important;
+    border: none !important;
+    font-weight: 600 !important;
+    font-size: 1.1em !important;
+    padding: 12px 24px !important;
+    transition: all 0.3s ease !important;
+}
+
+.gr-button-primary:hover {
+    transform: translateY(-2px) !important;
+    box-shadow: 0 8px 25px rgba(102, 126, 234, 0.4) !important;
+}
+
+.markdown-text {
+    color: #E0E0E0 !important;
+}
+
+.gr-box {
+    border-radius: 12px !important;
+    border: 1px solid rgba(255, 255, 255, 0.1) !important;
+}
+
+footer {
+    display: none !important;
+}
+"""
+
+with gr.Blocks(theme=gr.themes.Soft(), css=custom_css, title="SOHO Comet Detector") as demo:
     
-    **Upload a ZIP file** containing SOHO/LASCO C3 FITS images from a time sequence.
-    The AI will classify whether the sequence contains comet activity.
+    gr.HTML("""
+    <div style="text-align: center; margin-bottom: 2em;">
+        <h1>🌌 SOHO Comet Detection System</h1>
+        <p style="font-size: 1.2em; color: #B0B0B0; font-weight: 300;">
+            AI-Powered Analysis of NASA Solar Coronagraph Data
+        </p>
+    </div>
+    """)
     
-    ### 📋 Instructions:
-    1. Download SOHO LASCO C3 images (6-hour sequence recommended)
-    2. Put all .fts or .fits files in a folder
-    3. Zip the folder and upload here
-    4. Click "Analyze Sequence"
+    gr.Markdown("""
+    ### 🛸 About This System
+    
+    This deep learning system automatically detects sungrazing comets in SOHO/LASCO C3 
+    coronagraph images using difference imaging and convolutional neural networks.
     
-    ### 🔬 Model Details:
-    - **Architecture:** EfficientNet-B0
-    - **Method:** Difference imaging + full-image classification
-    - **Accuracy:** 97.7% on validation data
-    - **Precision:** 98% (comet detection)
-    - **Recall:** 99% (comet detection)
+    **Model Performance:**
+    - ✨ **97.7% Accuracy** on validation data
+    - 🎯 **98% Precision** for comet detection
+    - 🔍 **99% Recall** for comet detection
     
     ---
     """)
     
     with gr.Row():
-        with gr.Column():
+        with gr.Column(scale=1):
+            gr.Markdown("""
+            ### 📤 Upload Data
+            
+            **Instructions:**
+            1. Download SOHO LASCO C3 FITS images (6-hour time sequence recommended)
+            2. Place all `.fts` or `.fits` files in a folder
+            3. Create a ZIP archive of the folder
+            4. Upload the ZIP file below
+            5. Click "Analyze Sequence"
+            
+            **Maximum:** 50 images per upload
+            """)
+            
             zip_input = gr.File(
-                label="Upload ZIP of FITS Images", 
+                label="📁 Upload ZIP File", 
                 file_types=[".zip"], 
-                type="filepath"
+                type="filepath",
+                file_count="single"
             )
-            analyze_btn = gr.Button("🔍 Analyze Sequence", variant="primary", size="lg")
+            
+            analyze_btn = gr.Button(
+                "🔍 Analyze Sequence", 
+                variant="primary", 
+                size="lg"
+            )
+            
+            gr.Markdown("""
+            ---
+            
+            ### 🔬 Technical Details
+            
+            **Architecture:** EfficientNet-B0  
+            **Input Size:** 512×512 px  
+            **Method:** Maximum difference projection + binary classification  
+            **Training Data:** 498 comet + 167 background sequences
+            """)
         
-        with gr.Column():
-            output_plot = gr.Plot(label="Analysis Results")
-            output_text = gr.Markdown(label="Detection Summary")
+        with gr.Column(scale=2):
+            gr.Markdown("### 📊 Analysis Results")
+            
+            output_plot = gr.Plot(label="Visual Analysis")
+            output_text = gr.Markdown(label="Detection Report")
     
     gr.Markdown("""
     ---
-    ### 📊 About This System
     
-    This system uses difference imaging to highlight moving objects (comets) in SOHO coronagraph data,
-    then classifies the entire sequence using a convolutional neural network.
+    ### 📚 Data Source
+    
+    Training data sourced from the [NASA SOHO mission](https://soho.nascom.nasa.gov/) 
+    LASCO C3 coronagraph and the [Sungrazer Project](https://sungrazer.nrl.navy.mil/).
     
-    **Training Data:** 498 comet sequences + 167 background sequences from SOHO/LASCO C3
+    ### 👥 Development Team
     
-    **Team:** Sambhavi, Emily, Mohammed
+    **Sambhavi** • **Emily** • **Mohammed**
     
-    *Built with Gradio • Powered by Hugging Face*
+    ---
+    
+    *Built with Gradio • Powered by Hugging Face Spaces • 2025*
     """)
     
     analyze_btn.click(
@@ -183,4 +310,8 @@ with gr.Blocks(theme=gr.themes.Soft(), title="SOHO Comet Detector") as demo:
     )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        show_error=True
+    )