Top level not defined in button command Tkinter

Question

I added a top level in my tkinter GUI but I added a button which runs a deep learning model and is supposed to display the result in a separate label; however, when I run the code it says the top level is not defined.

def import_model(image):
    model = keras.models.load_model("C:\\Users\\Haame\\Documents\\AIIP\\my_model")
    image=cv2.imread(image)
    image = cv2.resize(image,(196,196))  
    predicted_label = model.predict(image[None,...]).argmax()

    #display result

    result = tk.Label(new, bg="black", fg="grey81", text=predicted_label, font=("Segoe Script", 25), borderwidth=3, relief="solid").pack()


def open_win(image):
    new= Toplevel(canvas)
    new.geometry("1000x850")
    new.title("New Window")
    Label(new, text="Selected image: ", font=('Helvetica 17 bold')).pack(pady=20)

    #create run button
    button2 = tk.Button (new, text='Run', font=("ROG FONTS", 10), bg='green', command = 
    import_model(image))
    button2.pack(pady=15)

I even tried passing the top level as a parameter in my function but it still did not work.

Edit: Full error

  File "C:\Users\Haame\Desktop\Project Front End.py", line 44, in import_model
    result = tk.Label(new, bg="black", fg="grey81", text=predicted_label, font=("Segoe Script", 25), borderwidth=3, relief="solid").pack()
NameError: name 'new' is not defined

Answer 1

Since new is a local variable inside open_win() , so it is not accessible inside import_model() . One of the way to fix it is to pass new as an argument to import_model() as well.

Note also that command=import_model(image) will execute import_model() immediately without clicking the button. lambda should be used in this case.

Below are the suggested changes:

# added new argument 'parent'
def import_model(parent, image):
    model = keras.models.load_model("C:\\Users\\Haame\\Documents\\AIIP\\my_model")
    image=cv2.imread(image)
    image = cv2.resize(image,(196,196))  
    predicted_label = model.predict(image[None,...]).argmax()

    #display result

    # used 'parent' argument
    tk.Label(parent, bg="black", fg="grey81", text=predicted_label, font=("Segoe Script", 25), borderwidth=3, relief="solid").pack()

def open_win(image):
    new= Toplevel(canvas)
    new.geometry("1000x850")
    new.title("New Window")
    Label(new, text="Selected image: ", font=('Helvetica 17 bold')).pack(pady=20)

    #create run button
    button2 = tk.Button (new, text='Run', font=("ROG FONTS", 10), bg='green',
                         # used lambda and pass 'new' to import_model() as well
                         command = lambda: import_model(new, image))
    button2.pack(pady=15)

---

Another suggestion is to create the result label inside open_win() and pass this label to import_model() :

def import_model(image, result):
    model = keras.models.load_model("C:\\Users\\Haame\\Documents\\AIIP\\my_model")
    image=cv2.imread(image)
    image = cv2.resize(image,(196,196))
    predicted_label = model.predict(image[None,...]).argmax()

    #display result
    result.config(text=predicted_label)
    result.pack()

def open_win(image):
    new= Toplevel(canvas)
    new.geometry("1000x850")
    new.title("New Window")
    Label(new, text="Selected image: ", font=('Helvetica 17 bold')).pack(pady=20)

    #create run button
    button2 = tk.Button (new, text='Run', font=("ROG FONTS", 10), bg='green',
                         command = lambda: import_model(image, result))
    button2.pack(pady=15)

    # create the result label and make it hidden initially
    result = tk.Label(new, bg="black", fg="grey81", font=("Segoe Script", 25), borderwidth=3, relief="solid")