update space

.idea/workspace.xml

@@ -5,10 +5,11 @@
   </component>
   <component name="ChangeListManager">
     <list default="true" id="27c9ae1a-a6fa-4472-8bcd-a7087620894b" name="Changes" comment="update space">
-      <change afterPath="$PROJECT_DIR$/training/resnet-18-model-for-poultry-disease-classification.ipynb" afterDir="false" />
+      <change afterPath="$PROJECT_DIR$/models/auth.h5" afterDir="false" />
       <change beforePath="$PROJECT_DIR$/.idea/workspace.xml" beforeDir="false" afterPath="$PROJECT_DIR$/.idea/workspace.xml" afterDir="false" />
-      <change beforePath="$PROJECT_DIR$/app.py" beforeDir="false" afterPath="$PROJECT_DIR$/app.py" afterDir="false" />
-      <change beforePath="$PROJECT_DIR$/requirements.txt" beforeDir="false" afterPath="$PROJECT_DIR$/requirements.txt" afterDir="false" />
+      <change beforePath="$PROJECT_DIR$/models/Final_Chicken_disease_model.h5" beforeDir="false" afterPath="$PROJECT_DIR$/models/diseases.h5" afterDir="false" />
+      <change beforePath="$PROJECT_DIR$/services/convert_model.py" beforeDir="false" afterPath="$PROJECT_DIR$/services/convert_model.py" afterDir="false" />
+      <change beforePath="$PROJECT_DIR$/services/disease_detection_service.py" beforeDir="false" afterPath="$PROJECT_DIR$/services/disease_detection_service.py" afterDir="false" />
     </list>
     <option name="SHOW_DIALOG" value="false" />
     <option name="HIGHLIGHT_CONFLICTS" value="true" />
@@ -51,7 +52,7 @@
     "RunOnceActivity.ShowReadmeOnStart": "true",
     "git-widget-placeholder": "main",
     "ignore.virus.scanning.warn.message": "true",
-    "last_opened_file_path": "C:/git/Generative_AI_with_poultry_disease_detection_system_v2",
+    "last_opened_file_path": "C:/git/Generative_AI_with_poultry_disease_detection_system_v2/models",
     "list.type.of.created.stylesheet": "CSS",
     "node.js.detected.package.eslint": "true",
     "node.js.detected.package.tslint": "true",
@@ -64,11 +65,11 @@
 }]]></component>
   <component name="RecentsManager">
     <key name="CopyFile.RECENT_KEYS">
+      <recent name="C:\git\Generative_AI_with_poultry_disease_detection_system_v2\models" />
       <recent name="C:\git\Generative_AI_with_poultry_disease_detection_system_v2" />
       <recent name="C:\git\Generative_AI_with_poultry_disease_detection_system_v2\static" />
       <recent name="C:\git\Generative_AI_with_poultry_disease_detection_system_v2\templates\tasks" />
       <recent name="C:\git\Generative_AI_with_poultry_disease_detection_system_v2\templates\todo" />
-      <recent name="C:\git\Generative_AI_with_poultry_disease_detection_system_v2\templates" />
     </key>
     <key name="MoveFile.RECENT_KEYS">
       <recent name="C:\git\Generative_AI_with_poultry_disease_detection_system_v2\static\images" />
@@ -130,14 +131,6 @@
       <workItem from="1730397485849" duration="22781000" />
       <workItem from="1730454506390" duration="12672000" />
     </task>
-    <task id="LOCAL-00075" summary="update space">
-      <option name="closed" value="true" />
-      <created>1730400686897</created>
-      <option name="number" value="00075" />
-      <option name="presentableId" value="LOCAL-00075" />
-      <option name="project" value="LOCAL" />
-      <updated>1730400686897</updated>
-    </task>
     <task id="LOCAL-00076" summary="update space">
       <option name="closed" value="true" />
       <created>1730402280098</created>
@@ -522,7 +515,15 @@
       <option name="project" value="LOCAL" />
       <updated>1730488911425</updated>
     </task>
-    <option name="localTasksCounter" value="124" />
+    <task id="LOCAL-00124" summary="update space">
+      <option name="closed" value="true" />
+      <created>1730490013545</created>
+      <option name="number" value="00124" />
+      <option name="presentableId" value="LOCAL-00124" />
+      <option name="project" value="LOCAL" />
+      <updated>1730490013545</updated>
+    </task>
+    <option name="localTasksCounter" value="125" />
     <servers />
   </component>
   <component name="TypeScriptGeneratedFilesManager">

models/auth.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:55ca5e07d8f4d45eab021364be749ced18402f85f5edb7425486ed76ea5c3093
+size 234256896

services/convert_model.py

@@ -2,7 +2,7 @@
 
 import tensorflow as tf
 
-MODEL_PATH = "models/Final_Chicken_disease_model.h5"
+MODEL_PATH = "models/diseases.h5"
 NEW_MODEL_PATH = "models/Final_Chicken_disease_model"
 
 # Load the .h5 model

services/disease_detection_service.py

@@ -3,19 +3,20 @@
 import os
 import logging
 import threading
-import torch
+import tensorflow as tf
 import cv2
 import numpy as np
+from keras.models import load_model
 from huggingface_hub import login
 from transformers import AutoTokenizer, AutoModelForCausalLM
 from services.utils import db  # MongoDB instance from utils
 import configparser
 
-# Configure logging with advanced settings for detailed output
+# Configure logging
 logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
 logger = logging.getLogger(__name__)
 
-# Load and configure Hugging Face API token for secure model access
+# Load Hugging Face API token for secure model access
 HF_TOKEN = os.environ.get('HF_Token')
 if HF_TOKEN:
     login(token=HF_TOKEN, add_to_git_credential=True)
@@ -23,44 +24,65 @@ if HF_TOKEN:
 else:
     logger.warning("Hugging Face token missing. Please set HF_Token in environment variables.")
 
-# Load configuration for model paths and other settings from 'config.ini'
+# Load configuration for model paths
 config = configparser.ConfigParser()
 config.read('config.ini')
 logger.info("Configuration loaded from config.ini.")
 
-# Check for GPU availability and set device
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-logger.info(f"Using device: {device}")
+# Configure TensorFlow for GPU and mixed precision if supported
+gpu_devices = tf.config.list_physical_devices('GPU')
+if gpu_devices:
+    for gpu in gpu_devices:
+        tf.config.experimental.set_memory_growth(gpu, True)
+    if tf.config.experimental.get_device_details(gpu_devices[0]).get('compute_capability', (0, 0))[0] >= 7:
+        from tensorflow.keras import mixed_precision
+        policy = mixed_precision.Policy('mixed_float16')
+        mixed_precision.set_global_policy(policy)
+        logger.info("Mixed precision enabled on supported GPU.")
+else:
+    logger.info("No GPU detected, using CPU without mixed precision.")
 
 # Global model variables and threading event for synchronization
-disease_model = None
+disease_model, auth_model = None, None
 llama_model, llama_tokenizer = None, None
 model_loading_event = threading.Event()
 
 # Function to load the disease detection model
 def load_disease_model():
-    """Load the pre-trained PyTorch model for poultry disease detection."""
+    """Load the disease detection model."""
     try:
-        model_path = config.get('MODELS', 'DISEASE_MODEL_PATH', fallback="models/Final_Chicken_disease_model.pth")
-        model = torch.load(model_path, map_location=device)
-        model.eval()  # Set model to evaluation mode
-        logger.info("Disease detection model loaded successfully.")
+        model_path = config.get('MODELS', 'DISEASE_MODEL_PATH', fallback="models/diseases.h5")
+        device = tf.config.list_logical_devices('GPU')[0] if gpu_devices else tf.config.list_logical_devices('CPU')[0]
+        with tf.device(device.name):
+            model = load_model(model_path, compile=True)
+        logger.info(f"Disease detection model loaded successfully on {device.name}.")
         return model
     except Exception as e:
         logger.error("Error loading disease detection model", exc_info=True)
         raise RuntimeError("Failed to load disease detection model") from e
 
+# Function to load the authentication model
+def load_auth_model():
+    """Load the authentication model to verify poultry images."""
+    try:
+        auth_model_path = config.get('MODELS', 'AUTH_MODEL_PATH', fallback="models/auth.h5")
+        auth_model = load_model(auth_model_path, compile=True)
+        logger.info("Authentication model loaded successfully.")
+        return auth_model
+    except Exception as e:
+        logger.error("Error loading authentication model", exc_info=True)
+        raise RuntimeError("Failed to load authentication model") from e
+
 # Function to load the Llama 3.2 model for language generation
 def load_llama_model():
     """Load the Llama model and tokenizer for language generation."""
     global llama_model, llama_tokenizer
     try:
         logger.info("Loading Llama 3.2 model and tokenizer using PyTorch-compatible model.")
-        model_name = "meta-llama/Llama-3.2-1B"  # Update with desired model name
+        model_name = "meta-llama/Llama-3.2-1B"
 
-        # Load the tokenizer and model from the Hugging Face hub
         llama_tokenizer = AutoTokenizer.from_pretrained(model_name)
-        llama_model = AutoModelForCausalLM.from_pretrained(model_name).to(device)
+        llama_model = AutoModelForCausalLM.from_pretrained(model_name)
 
         # Add a padding token if it's missing
         if llama_tokenizer.pad_token is None:
@@ -69,11 +91,11 @@ def load_llama_model():
             llama_model.resize_token_embeddings(len(llama_tokenizer))
 
         logger.info("Llama model and tokenizer loaded successfully.")
-        model_loading_event.set()  # Signal successful loading
+        model_loading_event.set()
         return llama_model, llama_tokenizer
     except Exception as e:
         logger.error("Failed to load Llama model", exc_info=True)
-        model_loading_event.set()  # Ensure event is set even if loading fails
+        model_loading_event.set()
         raise RuntimeError("Failed to load the Llama 3.2 model. Verify compatibility.")
 
 # Disease mappings and recommended treatments
@@ -86,63 +108,65 @@ recommendations = {
     3: 'Administer antibiotics as prescribed by a veterinarian and ensure biosecurity.'
 }
 
-class PoultryFarmBot:
-    """Poultry Farm Bot for disease prediction and response generation."""
 
-    def __init__(self, db):
-        """Initialize bot with MongoDB access for record-keeping."""
-        self.db = db
-        logger.info("PoultryFarmBot initialized with MongoDB access.")
+class PoultryFarmBot:
+    def __init__(self, disease_model, auth_model):
+        self.disease_model = disease_model
+        self.auth_model = auth_model
+        logging.basicConfig(level=logging.INFO)
+        self.logger = logging.getLogger(__name__)
 
-    def preprocess_image(self, image: np.ndarray) -> torch.Tensor:
+    def preprocess_image(self, image: np.ndarray) -> np.ndarray:
         """Preprocess the input image for model prediction."""
         try:
-            if image is None or image.size == 0:
-                raise ValueError("Invalid or empty image.")
             image = cv2.resize(image, (224, 224))
-            image = np.transpose(image, (2, 0, 1))  # Change from HWC to CHW format for PyTorch
-            image = torch.tensor(image, dtype=torch.float32).unsqueeze(0) / 255.0
-            logger.info("Image preprocessed successfully.")
-            return image.to(device)
+            return image / 255.0  # Normalize
         except Exception as e:
-            logger.error("Error in image preprocessing", exc_info=True)
-            raise
+            self.logger.error("Error in image preprocessing", exc_info=True)
+            raise ValueError("Invalid image format or empty image provided.")
+
+    def generate_detailed_response(self, disease_name: str, status: str, recommendation: str) -> str:
+        """Generate detailed response using Llama model based on prediction."""
+        prompt = (
+            f"The detected disease is {disease_name}, classified as {status}. "
+            f"Suggested action: {recommendation}. "
+            f"Here is additional information on {disease_name}: causes, symptoms, and effective management."
+        )
+        response = self.llama_response(prompt)
+        return response.replace(prompt, "").strip()
 
     def predict_disease(self, image: np.ndarray):
         """Predict disease from preprocessed image and provide detailed results."""
         try:
             preprocessed_image = self.preprocess_image(image)
-            with torch.no_grad():
-                prediction = disease_model(preprocessed_image)
-            predicted_class = torch.argmax(prediction, dim=1).item()
+
+            # Use auth_model to verify the image is poultry-related
+            is_poultry = self.auth_model.predict(preprocessed_image.reshape(1, 224, 224, 3)).argmax()
+            if is_poultry != 0:
+                self.logger.info("Image not recognized as poultry.")
+                return "Unknown Image", "N/A", "N/A", "Image not recognized as poultry.", None
+
+            # Predict disease if image is verified
+            prediction = self.disease_model.predict(preprocessed_image.reshape(1, 224, 224, 3))
+            predicted_class = prediction.argmax()
             disease_name = name_disease.get(predicted_class, "Unknown disease")
             disease_status = status_map.get(predicted_class, "Unknown status")
             recommendation = recommendations.get(predicted_class, "No recommendation available")
-            confidence = torch.softmax(prediction, dim=1)[0][predicted_class].item() * 100
-            logger.info(f"Disease Prediction: {disease_name} with {confidence:.2f}% confidence.")
+            confidence = prediction[0][predicted_class] * 100
+            self.logger.info(f"Disease Prediction: {disease_name} with {confidence:.2f}% confidence.")
 
             # Generate a detailed response using Llama model
             detailed_response = self.generate_detailed_response(disease_name, disease_status, recommendation)
             return detailed_response, disease_name, disease_status, recommendation, confidence
         except Exception as e:
-            logger.error("Error in disease prediction", exc_info=True)
+            self.logger.error("Error in disease prediction", exc_info=True)
             return "Prediction failed.", None, None, None, None
 
-    def generate_detailed_response(self, disease_name: str, status: str, recommendation: str) -> str:
-        """Generate detailed response using Llama model based on prediction."""
-        prompt = (
-            f"The detected disease is {disease_name}, classified as {status}. "
-            f"Suggested action: {recommendation}. "
-            f"Here is additional information on {disease_name}: causes, symptoms, and effective management."
-        )
-        response = self.llama_response(prompt)
-        return response.replace(prompt, "").strip()
-
     def llama_response(self, prompt: str) -> str:
         """Generate a response from the Llama model based on a prompt."""
         try:
             max_length = min(len(prompt) + 50, 512)
-            inputs = llama_tokenizer(prompt, return_tensors="pt", max_length=max_length, truncation=True, padding=True).to(device)
+            inputs = llama_tokenizer(prompt, return_tensors="tf", max_length=max_length, truncation=True, padding=True)
             outputs = llama_model.generate(inputs["input_ids"], max_length=max_length, do_sample=True, temperature=0.7, top_p=0.9)
             return llama_tokenizer.decode(outputs[0], skip_special_tokens=True)
         except Exception as e:
@@ -155,10 +179,12 @@ class PoultryFarmBot:
             return "Provide a valid poultry fecal image.", None, None, None, None
         return self.predict_disease(image)
 
+
 # Load models upon service startup
 disease_model = load_disease_model()
+auth_model = load_auth_model()
 llama_model, llama_tokenizer = load_llama_model()
 
 # Initialize the bot instance with MongoDB
-bot = PoultryFarmBot(db)
+bot = PoultryFarmBot(disease_model, auth_model)
 logger.info("Bot instance initialized with models loaded.")