import streamlit as st
import cv2
import numpy as np
from PIL import Image
from transformers import pipeline

# Function to load model from Hugging Face
@st.cache(allow_output_mutation=True)
def load_model():
    return pipeline("pose-detection", device=0)  # Adjust device as per your requirement

# Function to detect yoga pose from image
def detect_yoga_pose(image):
    # Convert PIL image to OpenCV format
    cv_image = cv2.cvtColor(np.array(image), cv2.COLOR_RGB2BGR)
    # Your pose detection logic here
    # Replace the following line with your actual pose detection code
    return "Detected yoga pose: Warrior II"

def main():
    st.title("Yoga Pose Detection from Live Camera Feed")

    # Load the model
    model = load_model()

    # Accessing the webcam
    cap = cv2.VideoCapture(0)

    # Run the app
    while True:
        ret, frame = cap.read()

        # Display the webcam feed
        st.image(frame, channels="BGR")

        # Convert the OpenCV frame to PIL image
        frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
        pil_image = Image.fromarray(frame)

        # Detect yoga pose from the image
        pose = detect_yoga_pose(pil_image)

        # Display the detected yoga pose
        st.write("Detected Yoga Pose:", pose)

        # Close the webcam feed
        if st.button("Stop"):
            break

    # Release the webcam and close Streamlit app
    cap.release()
    st.stop()

if __name__ == "__main__":
    main()