Upload 4 files

.gitattributes

@@ -35,3 +35,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 banana_disease.keras filter=lfs diff=lfs merge=lfs -text
 banana_disease_model.keras filter=lfs diff=lfs merge=lfs -text
+banana_disease_densenet121.keras filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,91 @@
+import gradio as gr
+import tensorflow as tf
+import numpy as np
+
+# Load the saved model
+model = tf.keras.models.load_model('banana_disease_densenet121.keras')
+
+# Load class names
+class_names = np.load("class_names.npy", allow_pickle=True)
+
+# Preprocess the input image
+def preprocess_image(img):
+    if img is None:
+        return None
+    img = img.resize((256, 256))  # Resize to match training
+    img = np.array(img) / 255.0   # Normalize
+    img = np.expand_dims(img, axis=0)  # Add batch dimension
+    return img
+
+# Prediction function
+def predict_disease(img):
+    img = preprocess_image(img)
+    if img is None:
+        return "⚠️ No image provided", None
+    predictions = model.predict(img)[0]
+    predicted_class = np.argmax(predictions)
+    confidence_scores = {class_names[i]: float(predictions[i]) for i in range(len(class_names))}
+    return f"Predicted: {class_names[predicted_class]}", confidence_scores
+
+
+# 🌿 Custom theme
+custom_theme = gr.themes.Soft()
+
+# UI
+with gr.Blocks(theme=custom_theme, css="""
+#title {
+  font-size: 32px;
+  font-weight: bold;
+  text-align: center;
+  color: white;
+  animation: fadeIn 2s ease-in-out;
+}
+#subtitle {
+  text-align: center;
+  font-size: 16px;
+  color: #ccc;
+  animation: fadeInUp 2s ease-in-out;
+}
+#prediction_box {
+  font-size: 18px;
+  padding: 12px;
+  border-radius: 8px;
+  min-height: 60px;   /* ensures readable area */
+  word-wrap: break-word;
+  white-space: normal;
+}
+.gr-button {
+  transition: 0.3s;
+}
+.gr-button:hover {
+  transform: scale(1.05);
+  box-shadow: 0px 4px 20px rgba(0, 128, 0, 0.4);
+}
+@keyframes fadeIn {
+  from {opacity: 0;}
+  to {opacity: 1;}
+}
+@keyframes fadeInUp {
+  from {opacity: 0; transform: translateY(10px);}
+  to {opacity: 1; transform: translateY(0);}
+}
+""") as demo:
+
+    gr.Markdown("<div id='title'>Banana Leaf Disease Classifier</div>")
+    gr.Markdown("<div id='subtitle'>Upload a banana leaf image, and our AI will diagnose the disease</div>")
+
+    with gr.Row():
+        with gr.Column(scale=1):
+            image_input = gr.Image(type="pil", image_mode="RGB", label="Upload Banana Leaf")
+            predict_btn = gr.Button("🔍 Predict Disease", variant="primary", elem_id="predict_btn")
+        
+        with gr.Column(scale=1):
+            # ✅ Changed from Textbox to Text
+            output_label = gr.Text(label="Prediction", interactive=False, elem_id="prediction_box")
+            with gr.Accordion("Confidence Scores", open=True):
+                output_conf = gr.Label(label="Scores")
+
+    predict_btn.click(fn=predict_disease, inputs=image_input, outputs=[output_label, output_conf])
+
+# Launch app
+demo.launch(share=True)

banana_disease.py

@@ -0,0 +1,132 @@
+import os
+import shutil
+import random
+import numpy as np
+import tensorflow as tf
+from tensorflow.keras import layers, models
+from tensorflow.keras.preprocessing.image import ImageDataGenerator
+from tensorflow.keras.callbacks import EarlyStopping, ReduceLROnPlateau
+from tensorflow.keras.applications.densenet import DenseNet121, preprocess_input
+
+# ---------------------------
+# Clear session
+# ---------------------------
+tf.keras.backend.clear_session()
+
+# ---------------------------
+# Paths
+# ---------------------------
+DATA_DIR = "/kaggle/input/Banana Disease Recognition Dataset/Original Images/Original Images"
+BASE_DIR = "/kaggle/working/banana_split"
+TRAIN_DIR = os.path.join(BASE_DIR, "train")
+VAL_DIR = os.path.join(BASE_DIR, "val")
+
+# ---------------------------
+# Create train/val split
+# ---------------------------
+os.makedirs(TRAIN_DIR, exist_ok=True)
+os.makedirs(VAL_DIR, exist_ok=True)
+
+for cls in os.listdir(DATA_DIR):
+    cls_path = os.path.join(DATA_DIR, cls)
+    if not os.path.isdir(cls_path):
+        continue
+    os.makedirs(os.path.join(TRAIN_DIR, cls), exist_ok=True)
+    os.makedirs(os.path.join(VAL_DIR, cls), exist_ok=True)
+    
+    files = [f for f in os.listdir(cls_path) if os.path.isfile(os.path.join(cls_path, f))]
+    random.shuffle(files)
+    split_idx = int(0.8 * len(files))
+    
+    for f in files[:split_idx]:
+        shutil.copy(os.path.join(cls_path, f), os.path.join(TRAIN_DIR, cls, f))
+    for f in files[split_idx:]:
+        shutil.copy(os.path.join(cls_path, f), os.path.join(VAL_DIR, cls, f))
+
+print("✅ Dataset successfully split into train & val folders")
+
+# ---------------------------
+# Parameters
+# ---------------------------
+IMG_SIZE = (256, 256)
+BATCH_SIZE = 32
+EPOCHS = 30
+
+# ---------------------------
+# Data Generators
+# ---------------------------
+train_datagen = ImageDataGenerator(
+    preprocessing_function=preprocess_input,
+    rotation_range=90,
+    horizontal_flip=True,
+    vertical_flip=True,
+    zoom_range=0.2
+)
+
+val_datagen = ImageDataGenerator(preprocessing_function=preprocess_input)
+
+train_generator = train_datagen.flow_from_directory(
+    TRAIN_DIR,
+    target_size=IMG_SIZE,
+    batch_size=BATCH_SIZE,
+    class_mode="categorical",
+    color_mode="rgb"
+)
+
+val_generator = val_datagen.flow_from_directory(
+    VAL_DIR,
+    target_size=IMG_SIZE,
+    batch_size=BATCH_SIZE,
+    class_mode="categorical",
+    color_mode="rgb"
+)
+
+# ---------------------------
+# Build model - DenseNet121
+# ---------------------------
+num_classes = train_generator.num_classes
+
+base_model = DenseNet121(
+    include_top=False,
+    weights='imagenet',
+    input_shape=(IMG_SIZE[0], IMG_SIZE[1], 3)
+)
+
+base_model.trainable = False  # Freeze initially
+
+x = layers.GlobalAveragePooling2D()(base_model.output)
+x = layers.Dropout(0.4)(x)
+output = layers.Dense(num_classes, activation='softmax')(x)
+
+model = models.Model(inputs=base_model.input, outputs=output)
+
+model.compile(
+    optimizer=tf.keras.optimizers.Adam(),
+    loss="categorical_crossentropy",
+    metrics=["accuracy"]
+)
+
+model.summary()
+
+# ---------------------------
+# Callbacks
+# ---------------------------
+early_stop = EarlyStopping(monitor="val_loss", patience=7, restore_best_weights=True, verbose=1)
+lr_reduce = ReduceLROnPlateau(monitor="val_loss", factor=0.2, patience=3, verbose=1)
+
+# ---------------------------
+# Train
+# ---------------------------
+history = model.fit(
+    train_generator,
+    validation_data=val_generator,
+    epochs=EPOCHS,
+    callbacks=[early_stop, lr_reduce]
+)
+
+# ---------------------------
+# Save class names & model
+# ---------------------------
+np.save("class_names.npy", np.array(list(train_generator.class_indices.keys())))
+model.save("banana_disease_densenet121.keras")
+print("✅ Training complete. Model saved as 'banana_disease_densenet121.keras'")

banana_disease_densenet121.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3667cd9cf42475cb6c81488380fcf47d4776e64e50c2dae807d7755476b41c9
+size 29746567

class_names.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8603247d4942390725f3051f812f316528334d181bcc757fb3a74e0f5a25b553
+size 1052