Fix: Embed model loading and prediction directly in main.py

main.py

@@ -5,18 +5,97 @@ from fastapi import FastAPI, File, UploadFile, HTTPException
 from fastapi.responses import JSONResponse
 from fastapi.middleware.cors import CORSMiddleware
 
-# Import the prediction function
+# Embedded prediction function with model loading
 import sys
-sys.path.append('/code')
+import numpy as np
+import tensorflow as tf
+from tensorflow.keras.preprocessing import image
+from huggingface_hub import hf_hub_download
 
-try:
-    from app.predict import predict_image
-    print("Successfully imported predict_image function")
-except ImportError as e:
-    print(f"Failed to import predict_image: {e}")
-    # Fallback prediction function for testing
-    def predict_image(image_path):
-        return "test_condition", 0.85, [{"label": "test_condition", "confidence": 0.85}]
+# Model configuration
+MODEL_PATH = "app/model/efficientnetv2s.h5"
+REPO_ID = "Miguel764/efficientnetv2s-skin-cancer-classifier"
+FILENAME = "efficientnetv2s.h5"
+TEMPERATURE = 2.77
+
+class_names_mapping = {
+    0: "AKIEC",
+    1: "BCC", 
+    2: "BKL",
+    3: "DF",
+    4: "MEL",
+    5: "NV",
+    6: "VASC"
+}
+
+full_names = {
+    "AKIEC": "Actinic keratoses and intraepithelial carcinoma",
+    "BCC": "Basal cell carcinoma",
+    "BKL": "Benign keratosis-like lesions", 
+    "DF": "Dermatofibroma",
+    "MEL": "Melanoma",
+    "NV": "Melanocytic nevi",
+    "VASC": "Vascular lesions"
+}
+
+# Global model variable
+model = None
+
+def load_model():
+    global model
+    try:
+        if not os.path.exists(MODEL_PATH):
+            print("Model not found locally. Downloading from Hugging Face...")
+            os.makedirs(os.path.dirname(MODEL_PATH), exist_ok=True)
+            hf_hub_download(
+                repo_id=REPO_ID,
+                filename=FILENAME,
+                local_dir="app/model"
+            )
+        else:
+            print("Model already exists locally.")
+        
+        print("Loading TensorFlow model...")
+        model = tf.keras.models.load_model(MODEL_PATH)
+        print("Model loaded successfully!")
+        return model
+    except Exception as e:
+        print(f"Error loading model: {e}")
+        return None
+
+def predict_image(image_path):
+    global model
+    try:
+        if model is None:
+            model = load_model()
+            if model is None:
+                return "Error: Model not loaded", 0.0, [{"label": "Error", "confidence": 0.0}]
+        
+        # Load and preprocess image
+        img = image.load_img(image_path, target_size=(224, 224))
+        img_array = image.img_to_array(img)
+        img_array = np.expand_dims(img_array, axis=0)
+        img_array = tf.keras.applications.imagenet_utils.preprocess_input(img_array)
+        
+        # Make prediction
+        logits = model.predict(img_array)
+        scaled_logits = logits / TEMPERATURE
+        scaled_probs = tf.nn.softmax(scaled_logits).numpy()[0]
+        
+        class_idx = int(np.argmax(scaled_probs))
+        top_label = full_names[class_names_mapping[class_idx]]
+        top_confidence = float(scaled_probs[class_idx])
+        
+        all_predictions = [
+            {"label": class_names_mapping[i], "confidence": float(pred)}
+            for i, pred in enumerate(scaled_probs)
+        ]
+        
+        return top_label, top_confidence, all_predictions
+        
+    except Exception as e:
+        print(f"Prediction error: {e}")
+        return f"Error: {str(e)}", 0.0, [{"label": "Error", "confidence": 0.0}]
 
 app = FastAPI(
     title="GP-Tea Skin Analysis API",