Update app.py

app.py

@@ -4,28 +4,24 @@ from tensorflow.keras.preprocessing import image
 import numpy as np
 from PIL import Image
 
-# Hugging Face model path for the filter model
-#FILTER_MODEL_PATH = "https://huggingface.co/spaces/Ten7/Maize_and_Cassava_Crop_Disease_Classification/tree/main/model_checkpoint"
-
-
-#Function to load the filter model directly
-#@st.cache_resource
-#def load_filter_model():
-#    model = load_model(FILTER_MODEL_PATH)
-#    return model
-
-# Load the filter model
-filter_model = load_model('model_checkpoint')
-
-# Load maize and cassava models
-corn_model = load_model('corn_model_mobilenetv2.h5')
-cassava_model = load_model('cassava_disease_model.h5')
-
-# Function to predict an uploaded image
-def predict_image(img, model, class_names):
+# Load the models
+filter_model = load_model("model_checkpoint")  # Replace with actual directory path
+corn_model = load_model("corn_model_mobilenetv2.h5")
+cassava_model = load_model("cassava_disease_model.h5")
+
+# Function to preprocess and predict with the filter model
+def is_leaf_image(img, model, threshold=0.5):
+    img_resized = img.resize((128, 128))  # Ensure correct size for the filter model
+    img_array = np.expand_dims(image.img_to_array(img_resized) / 255.0, axis=0)  # Normalize
+    prediction = model.predict(img_array)
+    leaf_probability = prediction[0][1]  # Assuming "Leaf" class is at index 1
+    return leaf_probability >= threshold, leaf_probability
+
+# Function to predict crop disease
+def predict_disease(img, model, class_names):
     img = image.img_to_array(img)
     img = np.expand_dims(img, axis=0)
-    img = img / 255.0  # Normalize image
+    img = img / 255.0  # Normalize
     prediction = model.predict(img)
     predicted_class = np.argmax(prediction)
     return predicted_class, class_names[predicted_class]
@@ -35,84 +31,66 @@ def home_page():
     st.title("Maize and Cassava Crop Disease Identification")
     st.image("1093908406.jpg", use_column_width=True)
     st.write("""
-        Welcome to the Crop Disease Identification App. This tool helps farmers identify diseases in maize and cassava crops by analyzing images of leaves.
-        
-        Use the navigation menu to explore the app:
-        - **Home:** Overview of the app
-        - **About:** Information about the app and its purpose
-        - **Prediction:** Upload an image and get a disease diagnosis along with recommendations.
+        Welcome to the Crop Disease Identification App. This tool aims at helping local farmers
+        identify diseases in maize and cassava crops by analyzing images of leaves.
     """)
 
 def about_page():
     st.title("About")
     st.image("IMG_20240727_143208_444.jpg", use_column_width=True)
     st.write("""
-        This application helps diagnose common diseases in maize and cassava crops. It uses deep learning models trained on thousands of leaf images to accurately identify diseases.
-        
-        Diseases include:
-        - Maize: Blight, Common Rust, Gray Leaf Spot
-        - Cassava: Bacterial Blight, Green Mottle, Brown Streak, Mosaic
-        
-        For questions or feedback, email: olorunnisholato7@gmail.com
+        This application is designed to assist farmers in diagnosing common diseases in maize and cassava crops.
+        The models were built using Convolutional Neural Networks (CNNs), trained on publicly available datasets.
     """)
 
 def prediction_page():
     st.title("Crop Disease Prediction")
+
+    # Select crop type
     crop_type = st.selectbox("Select the crop:", ["Maize", "Cassava"])
+
+    # File uploader
     uploaded_file = st.file_uploader("Upload an image of the leaf...", type=["jpg", "jpeg", "png"])
 
     if uploaded_file is not None:
+        # Load and display the uploaded image
         img = Image.open(uploaded_file)
+        st.image(img, caption="Uploaded Image", use_column_width=True)
 
-        # Set image size for filtering and crop-specific models
-        img_resized = img.resize((128, 128))  # For the filter model
-        img_array = np.expand_dims(image.img_to_array(img_resized) / 255.0, axis=0)
-
-        # Apply the filter model
-        filter_prediction = filter_model.predict(img_array)
-        filter_class = np.argmax(filter_prediction)
-        if filter_class == 0:  # Assuming 0 is "Non-Leaf"
-            st.warning("The uploaded image does not appear to contain a leaf. Please try again.")
-            return
-
-        # Resize for crop-specific models
-        target_size = (128, 128) if crop_type == "Maize" else (224, 224)
-        img = img.resize(target_size)
+        if st.button("Classify"):
+            st.write("Classifying...")
 
-        # Display the uploaded image
-        st.image(img, caption='Uploaded Image', use_column_width=True)
+            # Filter model prediction
+            is_leaf, leaf_probability = is_leaf_image(img, filter_model)
+            if not is_leaf:
+                st.warning(f"The uploaded image does not appear to contain a leaf. (Confidence: {leaf_probability:.2f})")
+                return
 
-        if st.button('Predict'):
-            st.write("Classifying...")
+            # Crop-specific model prediction
             if crop_type == "Maize":
-                class_names = ['Blight', 'Common_Rust', 'Gray_Leaf_Spot', 'Healthy']
-                predicted_class, class_name = predict_image(img, corn_model, class_names)
+                img_resized = img.resize((128, 128))
+                class_names = ["Blight", "Common_Rust", "Gray_Leaf_Spot", "Healthy"]
+                _, class_name = predict_disease(img_resized, corn_model, class_names)
             else:
-                class_names = [
-                    'Cassava_bacterial_blight',
-                    'Cassava_brown_streak_disease',
-                    'Cassava_green_mottle',
-                    'Cassava_healthy',
-                    'Cassava_mosaic_disease',
-                ]
-                predicted_class, class_name = predict_image(img, cassava_model, class_names)
+                img_resized = img.resize((224, 224))
+                class_names = ["Cassava_bacterial_blight", "Cassava_brown_streak_disease", "Cassava_green_mottle", "Cassava_healthy", "Cassava_mosaic_disease"]
+                _, class_name = predict_disease(img_resized, cassava_model, class_names)
 
+            # Display result
             st.write(f"Predicted Class: {class_name}")
-
-            # Recommendations based on predictions
-            if class_name in ['Healthy', 'Cassava_healthy']:
+            if "Healthy" in class_name:
                 st.success("The leaf is healthy. No action needed.")
             else:
                 recommendations = {
-                    'Blight': "Remove infected plants, apply fungicides, and rotate crops.",
-                    'Common_Rust': "Use resistant varieties and apply fungicides as needed.",
-                    'Gray_Leaf_Spot': "Ensure crop rotation, balanced nitrogen, and apply fungicides.",
-                    'Cassava_bacterial_blight': "Use disease-free planting material and remove affected plants.",
-                    'Cassava_brown_streak_disease': "Plant resistant varieties and manage whiteflies.",
-                    'Cassava_green_mottle': "Ensure good sanitation and use resistant varieties.",
-                    'Cassava_mosaic_disease': "Use resistant varieties and practice good field hygiene.",
+                    "Blight": "Recommendation: Remove infected plants and apply fungicides.",
+                    "Common_Rust": "Recommendation: Use resistant varieties and apply fungicides if necessary.",
+                    "Gray_Leaf_Spot": "Recommendation: Ensure proper crop rotation and apply fungicides.",
+                    "Cassava_bacterial_blight": "Recommendation: Remove infected plants and use disease-free planting material.",
+                    "Cassava_brown_streak_disease": "Recommendation: Plant resistant cassava varieties.",
+                    "Cassava_green_mottle": "Recommendation: Ensure good field sanitation.",
+                    "Cassava_mosaic_disease": "Recommendation: Use resistant varieties and practice good hygiene."
                 }
-                st.warning(recommendations[class_name])
+                st.warning(recommendations.get(class_name, "No specific recommendation available."))
 
 # Main app
 def main():
@@ -128,4 +106,3 @@ def main():
 
 if __name__ == "__main__":
     main()
-