Upload 3 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+trained_model.keras filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,208 @@
+import gradio as gr
+import cv2
+import tensorflow as tf
+import numpy as np
+from collections import deque
+
+# -----------------------------
+# Load Model (Global)
+# -----------------------------
+print("Loading model...")
+model = tf.keras.models.load_model("trained_model.keras")
+print("Model loaded.")
+
+# -----------------------------
+# Class Labels
+# -----------------------------
+class_names = [
+    'Apple___Apple_scab', 'Apple___Black_rot', 'Apple___Cedar_apple_rust', 'Apple___healthy',
+    'Blueberry___healthy',
+    'Cherry_(including_sour)___Powdery_mildew', 'Cherry_(including_sour)___healthy',
+    'Corn_(maize)___Cercospora_leaf_spot Gray_leaf_spot', 'Corn_(maize)___Common_rust_',
+    'Corn_(maize)___Northern_Leaf_Blight', 'Corn_(maize)___healthy',
+    'Grape___Black_rot', 'Grape___Esca_(Black_Measles)', 'Grape___Leaf_blight_(Isariopsis_Leaf_Spot)', 'Grape___healthy',
+    'Orange___Haunglongbing_(Citrus_greening)',
+    'Peach___Bacterial_spot', 'Peach___healthy',
+    'Pepper,_bell___Bacterial_spot', 'Pepper,_bell___healthy',
+    'Potato___Early_blight', 'Potato___Late_blight', 'Potato___healthy',
+    'Raspberry___healthy',
+    'Soybean___healthy',
+    'Squash___Powdery_mildew',
+    'Strawberry___Leaf_scorch', 'Strawberry___healthy',
+    'Tomato___Bacterial_spot', 'Tomato___Early_blight', 'Tomato___Late_blight',
+    'Tomato___Leaf_Mold', 'Tomato___Septoria_leaf_spot',
+    'Tomato___Spider_mites Two-spotted_spider_mite', 'Tomato___Target_Spot',
+    'Tomato___Tomato_Yellow_Leaf_Curl_Virus', 'Tomato___Tomato_mosaic_virus', 'Tomato___healthy'
+]
+
+# -----------------------------
+# Global state for stabilization (for streaming only)
+# -----------------------------
+history = deque(maxlen=5) 
+
+# -----------------------------
+# Preprocessing Function (CRITICAL FIX)
+# -----------------------------
+def preprocess_frame(frame):
+    """
+    Handles any input frame (RGB, RGBA, Grayscale) from Gradio and
+    converts it to the exact (64, 64) BGR format your model was trained on.
+    """
+    
+    # --- 1. Robustly convert to 3-channel RGB ---
+    if len(frame.shape) == 2:
+        # It's Grayscale (H, W)
+        frame_rgb = cv2.cvtColor(frame, cv2.COLOR_GRAY2RGB)
+    elif frame.shape[2] == 1:
+        # It's Grayscale (H, W, 1)
+        frame_rgb = cv2.cvtColor(frame, cv2.COLOR_GRAY2RGB)
+    elif frame.shape[2] == 4:
+        # It's RGBA (H, W, 4)
+        frame_rgb = cv2.cvtColor(frame, cv2.COLOR_RGBA2RGB)
+    else:
+        # It's already 3-channel RGB (H, W, 3)
+        frame_rgb = frame
+
+    # --- 2. Convert from RGB to BGR (as model expects) ---
+    # Your original script used cv2.VideoCapture, which provides BGR frames.
+    frame_bgr = cv2.cvtColor(frame_rgb, cv2.COLOR_RGB2BGR)
+
+    # --- 3. Resize to model's input size (64, 64) ---
+    # Your original script used (64, 64).
+    img_resized = cv2.resize(frame_bgr, (64, 64))
+
+    # --- 4. Normalize and add batch dimension ---
+    img_normalized = img_resized / 255.0
+    img_normalized = img_normalized.astype(np.float32)
+    img_batch = np.expand_dims(img_normalized, axis=0)
+    
+    return img_batch
+
+# -----------------------------
+# Prediction Function for STREAMING (Webcam)
+# -----------------------------
+def predict_stream(frame):
+    """
+    Takes a single RGB frame, predicts, and returns an annotated RGB frame.
+    Uses 'history' deque for stabilization.
+    """
+    if frame is None:
+        return None
+    
+    # 1. Preprocess and predict
+    # (preprocess_frame now handles RGB -> BGR conversion)
+    preprocessed_img = preprocess_frame(frame)
+    prediction = model.predict(preprocessed_img, verbose=0)
+    predicted_class = np.argmax(prediction)
+    history.append(predicted_class)
+
+    # 2. Stabilize prediction
+    if len(history) > 0:
+        label_index = max(set(history), key=history.count)
+        label = class_names[label_index]
+    else:
+        label = "Initializing..."
+
+    # 3. Return the original frame and the label text
+    # (All cv2.putText and color conversion logic removed)
+    return frame, label
+
+# -----------------------------
+# Prediction Function for UPLOAD (Single Image)
+# -----------------------------
+def predict_upload(frame):
+    """
+    Takes a single RGB frame, predicts, and returns an annotated RGB frame.
+    Does NOT use 'history' deque.
+    """
+    if frame is None:
+        return None
+    
+    # 1. Preprocess and predict
+    # (preprocess_frame now handles RGB -> BGR conversion)
+    preprocessed_img = preprocess_frame(frame)
+    prediction = model.predict(preprocessed_img, verbose=0)
+    predicted_class = np.argmax(prediction)
+    
+    # 2. Get label (no stabilization needed)
+    label = class_names[predicted_class]
+
+    # 3. Robustly convert original frame to RGB for display
+    # (This logic is still needed so the output image displays correctly)
+    if len(frame.shape) == 2:
+        frame_rgb = cv2.cvtColor(frame, cv2.COLOR_GRAY2RGB)
+    elif frame.shape[2] == 1:
+        frame_rgb = cv2.cvtColor(frame, cv2.COLOR_GRAY2RGB)
+    elif frame.shape[2] == 4:
+        frame_rgb = cv2.cvtColor(frame, cv2.COLOR_RGBA2RGB)
+    else:
+        frame_rgb = frame.copy()
+
+    # 4. Return the RGB frame and the label text
+    # (All cv2.putText and color conversion logic removed)
+    return frame_rgb, label
+
+# -----------------------------
+# Gradio Interface (with Tabs)
+# -----------------------------
+
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown(
+        """
+        # 🌱 Real-Time Plant Disease Detection
+        This app uses a trained CNN to detect plant diseases.
+        Use the tabs below to either start a live webcam feed or upload an image.
+        """
+    )
+    
+    with gr.Tabs():
+        # --- Tab 1: Live Detection ---
+        with gr.TabItem("Live Detection"):
+            with gr.Row():
+                webcam_input = gr.Image(
+                    sources=["webcam"],
+                    streaming=True,
+                    label="Webcam Feed"
+                )
+                webcam_output = gr.Image(label="Prediction")
+            
+            # --- NEW: Add a Label output for the prediction ---
+            webcam_label = gr.Label(label="Result")
+            
+            webcam_input.stream(
+                predict_stream, 
+                webcam_input, 
+                [webcam_output, webcam_label] # --- UPDATED: Output to both components ---
+            )
+
+        # --- Tab 2: Upload Image ---
+        with gr.TabItem("Upload Image"):
+            with gr.Row():
+                upload_input = gr.Image(
+                    sources=["upload"],
+                    label="Upload a plant image",
+                    type="numpy"
+                )
+                upload_output = gr.Image(label="Prediction")
+            
+            # --- NEW: Add a Label output for the prediction ---
+            upload_label = gr.Label(label="Result")
+
+            upload_input.upload(
+                predict_upload, 
+                upload_input, 
+                [upload_output, upload_label] # --- UPDATED: Output to both components ---
+            )
+            
+    # --- Accordions for extra info ---
+    with gr.Accordion("About this App"):
+        gr.Markdown("This project uses a TensorFlow/Keras CNN model to classify 38 different plant disease categories in real-time.")
+    
+    with gr.Accordion("Show all 38 classes"):
+        gr.JSON(class_names) 
+
+# -----------------------------
+# Launch the App
+# -----------------------------
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,9 @@
+gradio
+scikit-learn
+matplotlib
+seaborn
+pandas
+librosa
+tensorflow
+opencv-python-headless
+numpy

trained_model.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b2762962ee4a591eef3013445a26c7d04d27a28620ef05908f1122d80bdc13c
+size 10201218