model.py

from roboflow import Roboflow
import streamlit as st

def model():
    st.title('ASL Detection :mag:')

    # Roboflow integration and ASL detection code
    rf = Roboflow(api_key="07xd4Fb9Jmy0RHi57x5x")
    project = rf.workspace().project("american-sign-language-letters")
    model = project.version(6).model

    # predict function
    def predict(image_path):
        response = model.predict(image_path, confidence=40, overlap=30).json()

        if 'predictions' in response:
            predictions = response['predictions']
            if predictions:
                first_prediction = predictions[0]
                if 'class' in first_prediction:
                    return first_prediction['class']

        return None
    
    user_input = st.text_input("Enter your input").upper()
    picture = st.camera_input("Take a picture")

    if picture:
        with open("captured_image.jpg", "wb") as f:
            f.write(picture.getbuffer())
        st.write('Image captured and saved as captured_image.jpg')

        result = predict("captured_image.jpg")

        if result == None:
            st.write("Sign unclear, keep practicing!")
        elif result == user_input: 
            st.write(f"You signed {user_input} correctly!")
        else:
            st.write(f"Try again. You signed {user_input}, and we detected {result}")