update space

.env

@@ -10,5 +10,6 @@ SMTP_PASSWORD="your_email_password"
 FROM_EMAIL="your_from_email_address"
 
 
-# TensorFlow Optimization
+# TensorFlow Settings
 TF_ENABLE_ONEDNN_OPTS=0
+TF_FORCE_GPU_ALLOW_GROWTH=true

Dockerfile

@@ -1,4 +1,4 @@
-# Use the official Python 3.10 image
+# Use the official Python 3.9 image
 FROM python:3.9
 
 # Create and set up a user

app.py

@@ -11,6 +11,7 @@ import tensorflow as tf
 from routes.authentication import auth_router
 from routes.disease_detection import disease_router
 from routes.health_dashboard import dashboard_router
+from services.disease_detection_service import load_disease_model, load_llama_model
 from services.health_monitoring_service import evaluate_health_data, get_health_alerts, send_alerts
 from huggingface_hub import login
 
@@ -44,7 +45,9 @@ if os.path.isdir(static_dir):
 else:
     logger.error("Static directory not found.")
     raise HTTPException(status_code=500, detail="Static directory not found.")
-
+# Load models at startup
+disease_model = load_disease_model()
+llama_model, llama_tokenizer = load_llama_model()
 # Include routers for authentication, disease detection, and health dashboard
 app.include_router(auth_router, prefix="/auth", tags=["Authentication"])
 app.include_router(disease_router, prefix="/disease", tags=["Disease Detection"])

requirements.txt

@@ -1,4 +1,4 @@
-opencv-python
+opencv-python-headless
 fastapi
 passlib[bcrypt]
 pydantic[email]

services/disease_detection_service.py

@@ -38,24 +38,39 @@ if gpus:
 else:
     logger.info("Using CPU without mixed precision.")
 
-# Load the disease detection model
-try:
-    device_name = '/GPU:0' if gpus else '/CPU:0'
-    with tf.device(device_name):
-        disease_model = load_model('models/Final_Chicken_disease_model.h5', compile=True)
-    logger.info(f"Disease detection model loaded on {device_name}.")
-except Exception as e:
-    logger.error(f"Error loading disease model: {e}")
-
-# Llama 3.2 setup
-model_name = "meta-llama/Llama-3.2-1B"
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-llama_model = AutoModelForCausalLM.from_pretrained(model_name)
-
-# Set a padding token if it doesn’t already exist
-if tokenizer.pad_token is None:
-    tokenizer.add_special_tokens({'pad_token': '[PAD]'})
-    llama_model.resize_token_embeddings(len(tokenizer))
+
+# Model loading functions
+def load_disease_model():
+    """Load the disease detection model."""
+    try:
+        model_path = "models/Final_Chicken_disease_model.h5"
+        device = '/GPU:0' if gpu_devices else '/CPU:0'
+        with tf.device(device):
+            model = load_model(model_path, compile=True)
+        logger.info(f"Disease detection model loaded successfully on {device}.")
+        return model
+    except Exception as e:
+        logger.error(f"Error loading disease model: {e}")
+        raise RuntimeError("Failed to load disease detection model")
+
+
+def load_llama_model():
+    """Load the Llama 3.2 model for text generation."""
+    try:
+        model_name = "meta-llama/Llama-3.2-1B"
+        tokenizer = AutoTokenizer.from_pretrained(model_name)
+        model = AutoModelForCausalLM.from_pretrained(model_name)
+
+        if tokenizer.pad_token is None:
+            tokenizer.add_special_tokens({'pad_token': '[PAD]'})
+            model.resize_token_embeddings(len(tokenizer))
+
+        logger.info("Llama 3.2 model loaded successfully.")
+        return model, tokenizer
+    except Exception as e:
+        logger.error(f"Error loading Llama model: {e}")
+        raise RuntimeError("Failed to load Llama model")
+
 
 # Disease mapping and treatment guidelines
 name_disease = {0: 'Coccidiosis', 1: 'Healthy', 2: 'New Castle Disease', 3: 'Salmonella'}