Update app.py

app.py

@@ -5,44 +5,62 @@ from PIL import Image
 import json
 
 # ============================
-# LOAD MODEL + LABELS
+# LOAD MODEL (.h5)
 # ============================
-model = tf.keras.models.load_model("lemon_model.h5")
+model = tf.keras.models.load_model("lemon_model.h5", compile=False)
 
+# Recompiler (important pour certains .h5)
+model.compile(
+    optimizer="adam",
+    loss="categorical_crossentropy",
+    metrics=["accuracy"]
+)
+
+# ============================
+# LOAD LABELS
+# ============================
 with open("labels.json", "r") as f:
     class_names = json.load(f)
 
 IMG_SIZE = (224, 224)
 
 # ============================
-# PREDICTION FUNCTION
+# PREPROCESS
 # ============================
-def predict(image):
-    image = image.resize(IMG_SIZE)
-    img_array = np.array(image) / 255.0
+def preprocess(img):
+    img = img.convert("RGB")
+    img = img.resize(IMG_SIZE)
+    img_array = np.array(img) / 255.0
     img_array = np.expand_dims(img_array, axis=0)
+    return img_array
 
-    predictions = model.predict(img_array)[0]
+# ============================
+# PREDICTION
+# ============================
+def predict(img):
+    img_array = preprocess(img)
+    preds = model.predict(img_array)[0]
 
-    results = {
-        class_names[i]: float(predictions[i])
+    result = {
+        class_names[i]: float(preds[i])
         for i in range(len(class_names))
     }
 
-    return results
+    return result
 
 # ============================
 # INTERFACE
 # ============================
-interface = gr.Interface(
+demo = gr.Interface(
     fn=predict,
     inputs=gr.Image(type="pil"),
     outputs=gr.Label(num_top_classes=3),
-    title="🍋 Lemon Leaf Disease Detection",
-    description="Upload an image of a lemon leaf to detect disease"
+    title="🍋 Lemon Leaf Disease Detection (H5 Model)",
+    description="Upload a lemon leaf image to detect disease"
 )
 
 # ============================
 # LAUNCH
 # ============================
-interface.launch()
+if __name__ == "__main__":
+    demo.launch()