Simplify app structure - remove FastAPI conflict

app.py

@@ -6,8 +6,6 @@ import base64
 from io import BytesIO
 import json
 from huggingface_hub import hf_hub_download
-from fastapi import FastAPI, Request
-from fastapi.responses import JSONResponse
 
 print("🚀 Starting Affecto Inference Service...")
 
@@ -29,7 +27,7 @@ print(f"✅ Model downloaded to: {model_path}")
 
 # Load checkpoint
 checkpoint = torch.load(model_path, map_location=device)
-print(f"📦 Checkpoint keys: {checkpoint.keys()}")
+print(f"📦 Checkpoint loaded successfully")
 
 # ============================================
 # IMAGE PROCESSING UTILITIES
@@ -42,24 +40,22 @@ def preprocess_image(image):
     if not isinstance(image, Image.Image):
         image = Image.fromarray(image)
     
-    # Ensure RGB
     if image.mode != 'RGB':
         image = image.convert('RGB')
     
-    # Transform
     transform = transforms.Compose([
         transforms.Resize((256, 256)),
         transforms.ToTensor(),
         transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
     ])
     
-    tensor = transform(image).unsqueeze(0)  # Add batch dimension
+    tensor = transform(image).unsqueeze(0)
     return tensor.to(device)
 
 def postprocess_tensor(tensor):
     """Convert tensor to PIL image"""
     tensor = tensor.squeeze(0).cpu()
-    tensor = tensor * 0.5 + 0.5  # Denormalize
+    tensor = tensor * 0.5 + 0.5
     tensor = torch.clamp(tensor, 0, 1)
     
     numpy_image = tensor.numpy().transpose(1, 2, 0)
@@ -79,59 +75,44 @@ def base64_to_pil(base64_str):
     return Image.open(BytesIO(image_bytes))
 
 # ============================================
-# MOCK INFERENCE (Replace with actual model)
+# TRANSFORMATION
 # ============================================
 
 def apply_emotion_transform(input_tensor, au_params):
-    """
-    Apply emotion transformation
-    
-    TODO: Replace this with actual MagicFace inference when you have the model architecture
-    For now, this is a placeholder that applies simple image adjustments
-    """
+    """Apply emotion transformation (placeholder)"""
     print(f"🎭 Applying transformation with AU params: {au_params}")
     
-    # PLACEHOLDER: Simple brightness/contrast adjustment based on AU params
-    # In production, replace this with actual model inference
     output = input_tensor.clone()
     
-    # Example: Adjust based on AU12 (smile)
     if "AU12" in au_params:
         intensity = au_params["AU12"]
-        output = output * (1.0 + intensity * 0.2)  # Brighten
+        output = output * (1.0 + intensity * 0.2)
     
-    # Example: Adjust based on AU4 (frown)
     if "AU4" in au_params:
         intensity = au_params["AU4"]
-        output = output * (1.0 - intensity * 0.15)  # Darken
+        output = output * (1.0 - intensity * 0.15)
     
     output = torch.clamp(output, -1, 1)
-    
     return output
 
 # ============================================
-# API ENDPOINT (For Flutter App)
+# API FUNCTIONS
 # ============================================
 
-def transform_api(image_base64, au_params):
+def transform_api(data):
     """API function for external calls"""
     try:
+        image_base64 = data["image"]
+        au_params = data["au_params"]
+        
         print(f"📥 Received API request with AU params: {au_params}")
         
-        # Decode image
         image = base64_to_pil(image_base64)
         print(f"📸 Image size: {image.size}")
         
-        # Preprocess
         input_tensor = preprocess_image(image)
-        
-        # Transform
         output_tensor = apply_emotion_transform(input_tensor, au_params)
-        
-        # Postprocess
         result_image = postprocess_tensor(output_tensor)
-        
-        # Encode result
         result_base64 = pil_to_base64(result_image)
         
         print("✅ Transformation complete")
@@ -152,25 +133,38 @@ def transform_api(image_base64, au_params):
             "message": "Transformation failed"
         }
 
+def health_check():
+    """Health check function"""
+    return {
+        "status": "healthy",
+        "model": "magicface",
+        "device": str(device),
+        "version": "1.0.0"
+    }
+
+def root_info():
+    """Root info function"""
+    return {
+        "message": "Affecto Inference API",
+        "status": "running",
+        "version": "1.0.0",
+        "endpoints": {
+            "health": "/health",
+            "transform": "/transform"
+        }
+    }
+
 # ============================================
-# GRADIO INTERFACE (For Testing)
+# GRADIO INTERFACE
 # ============================================
 
 def transform_gradio(image, au_params_str):
     """Gradio interface function"""
     try:
-        # Parse AU params
         au_params = json.loads(au_params_str)
-        
-        # Preprocess
         input_tensor = preprocess_image(image)
-        
-        # Transform
         output_tensor = apply_emotion_transform(input_tensor, au_params)
-        
-        # Postprocess
         result_image = postprocess_tensor(output_tensor)
-        
         return result_image
     except Exception as e:
         print(f"❌ Error: {str(e)}")
@@ -178,59 +172,7 @@ def transform_gradio(image, au_params_str):
         traceback.print_exc()
         return image
 
-# ============================================
-# FASTAPI APP (Must be created BEFORE Gradio)
-# ============================================
-
-app = FastAPI(title="Affecto Inference API")
-
-@app.get("/")
-async def root():
-    """Root endpoint"""
-    return {
-        "message": "Affecto Inference API",
-        "status": "running",
-        "version": "1.0.0",
-        "endpoints": {
-            "health": "/health",
-            "transform": "/transform",
-            "gradio_ui": "/gradio"
-        }
-    }
-
-@app.get("/health")
-async def health_check():
-    """Health check endpoint"""
-    return {
-        "status": "healthy",
-        "model": "magicface",
-        "device": str(device),
-        "version": "1.0.0"
-    }
-
-@app.post("/transform")
-async def api_transform_endpoint(request: Request):
-    """Main transformation endpoint"""
-    try:
-        data = await request.json()
-        result = transform_api(
-            image_base64=data["image"],
-            au_params=data["au_params"]
-        )
-        return JSONResponse(content=result)
-    except Exception as e:
-        return JSONResponse(
-            content={
-                "success": False,
-                "error": str(e)
-            },
-            status_code=500
-        )
-
-# ============================================
-# GRADIO UI
-# ============================================
-
+# Build Gradio interface
 with gr.Blocks(theme=gr.themes.Soft(), title="Affecto Inference API") as demo:
     gr.Markdown("# 🎭 Affecto - Emotion Transformation API")
     gr.Markdown("Transform facial emotions using MagicFace Action Units")
@@ -250,17 +192,15 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Affecto Inference API") as demo:
                 output_image = gr.Image(type="pil", label="Transformed Result")
         
         gr.Markdown("### 🎨 Emotion Presets:")
-        with gr.Row():
-            gr.Examples(
-                examples=[
-                    ['{"AU6": 1.0, "AU12": 1.0}', "😊 Happy"],
-                    ['{"AU1": 1.0, "AU4": 1.0, "AU15": 1.0}', "😢 Sad"],
-                    ['{"AU4": 1.0, "AU5": 1.0, "AU7": 1.0, "AU23": 1.0}', "😠 Angry"],
-                    ['{"AU1": 1.0, "AU2": 1.0, "AU5": 1.0, "AU26": 1.0}', "😮 Surprised"],
-                ],
-                inputs=[au_params_input],
-                label="Click to use preset"
-            )
+        gr.Examples(
+            examples=[
+                ['{"AU6": 1.0, "AU12": 1.0}'],
+                ['{"AU1": 1.0, "AU4": 1.0, "AU15": 1.0}'],
+                ['{"AU4": 1.0, "AU5": 1.0, "AU7": 1.0, "AU23": 1.0}'],
+                ['{"AU1": 1.0, "AU2": 1.0, "AU5": 1.0, "AU26": 1.0}'],
+            ],
+            inputs=[au_params_input],
+        )
         
         transform_btn.click(
             fn=transform_gradio,
@@ -272,91 +212,58 @@ with gr.Blocks(theme=gr.themes.Soft(), title="Affecto Inference API") as demo:
         gr.Markdown("""
         ## API Endpoints
         
-        ### 1. Health Check
-        **GET** `/health`
-        
-        ```bash
-        curl https://gauravvjhaa-affecto-inference.hf.space/health
-        ```
-        
-        ### 2. Transform Image
-        **POST** `/transform`
+        ### Transform Image
+        **POST** `/api/transform`
         
-        **Request Format:**
         ```json
         {
-            "image": "base64_encoded_image_string",
-            "au_params": {
-                "AU6": 1.0,
-                "AU12": 1.0
-            }
-        }
-        ```
-        
-        **Response Format:**
-        ```json
-        {
-            "success": true,
-            "transformed_image": "base64_encoded_result",
-            "au_params": {...},
-            "message": "Transformation successful"
+            "image": "base64_encoded_image",
+            "au_params": {"AU6": 1.0, "AU12": 1.0}
         }
         ```
         
-        ### Available Action Units (AU):
-        - **AU1**: Inner Brow Raiser
-        - **AU2**: Outer Brow Raiser
-        - **AU4**: Brow Lowerer
-        - **AU5**: Upper Lid Raiser
-        - **AU6**: Cheek Raiser
-        - **AU7**: Lid Tightener
-        - **AU9**: Nose Wrinkler
-        - **AU12**: Lip Corner Puller (Smile)
-        - **AU15**: Lip Corner Depressor
-        - **AU17**: Chin Raiser
-        - **AU20**: Lip Stretcher
-        - **AU23**: Lip Tightener
-        - **AU25**: Lips Part
-        - **AU26**: Jaw Drop
-        
-        ### Example Usage (Python):
-        ```python
-        import requests
-        import base64
-        
-        # Read image
-        with open("image.jpg", "rb") as f:
-            image_base64 = base64.b64encode(f.read()).decode()
+        ### Health Check
+        **GET** `/api/health`
         
-        # Make request
-        response = requests.post(
-            "https://gauravvjhaa-affecto-inference.hf.space/transform",
-            json={
-                "image": image_base64,
-                "au_params": {"AU6": 1.0, "AU12": 1.0}
-            }
-        )
-        
-        result = response.json()
-        print(result["success"])
-        ```
-        
-        ### Example Usage (cURL):
-        ```bash
-        curl -X POST https://gauravvjhaa-affecto-inference.hf.space/transform \\
-          -H "Content-Type: application/json" \\
-          -d '{
-            "image": "YOUR_BASE64_IMAGE",
-            "au_params": {"AU6": 1.0, "AU12": 1.0}
-          }'
-        ```
+        Returns service status and model information.
         """)
+    
+    # API endpoints as Gradio functions
+    with gr.Tab("🔌 API"):
+        with gr.Row():
+            with gr.Column():
+                gr.Markdown("### POST /api/transform")
+                api_input = gr.Textbox(
+                    label="Request JSON",
+                    value='{"image": "BASE64_STRING", "au_params": {"AU6": 1.0}}',
+                    lines=5
+                )
+                api_btn = gr.Button("Test API")
+                api_output = gr.JSON(label="Response")
+                
+                api_btn.click(
+                    fn=lambda x: transform_api(json.loads(x)),
+                    inputs=[api_input],
+                    outputs=[api_output]
+                )
+            
+            with gr.Column():
+                gr.Markdown("### GET /api/health")
+                health_btn = gr.Button("Check Health")
+                health_output = gr.JSON(label="Health Status")
+                
+                health_btn.click(
+                    fn=health_check,
+                    inputs=[],
+                    outputs=[health_output]
+                )
 
-# Mount Gradio to FastAPI at /gradio path
-app = gr.mount_gradio_app(app, demo, path="/gradio")
+# Add API routes using Gradio's API
+demo.api_names = ["transform", "health", "root"]
 
 print("✅ Affecto Inference API Ready!")
-print(f"🌐 Root: https://gauravvjhaa-affecto-inference.hf.space/")
-print(f"🎨 Gradio UI: https://gauravvjhaa-affecto-inference.hf.space/gradio")
-print(f"🔌 API Health: https://gauravvjhaa-affecto-inference.hf.space/health")
-print(f"🔌 API Transform: https://gauravvjhaa-affecto-inference.hf.space/transform")
+print(f"🌐 Gradio UI: https://gauravvjhaa-affecto-inference.hf.space/")
+
+# Launch
+if __name__ == "__main__":
+    demo.launch(server_name="0.0.0.0", server_port=7860)