pages/1_📷️_Live_Stream.py

import av
import os
import sys
import streamlit as st
from streamlit_webrtc import VideoHTMLAttributes, webrtc_streamer
from aiortc.contrib.media import MediaRecorder


BASE_DIR = os.path.abspath(os.path.join(__file__, '../../'))
sys.path.append(BASE_DIR)


from utils import get_mediapipe_pose
from process_frame import ProcessFrame
from thresholds import get_thresholds_beginner, get_thresholds_pro


st.title('Anju AI Fitness Trainer')

mode = st.radio('Select Mode', ['Beginner', 'Pro'], horizontal=True)

thresholds = None 

if mode == 'Beginner':
    thresholds = get_thresholds_beginner()

elif mode == 'Pro':
    thresholds = get_thresholds_pro()


live_process_frame = ProcessFrame(thresholds=thresholds, flip_frame=True)
# Initialize face mesh solution
pose = get_mediapipe_pose()


if 'download' not in st.session_state:
    st.session_state['download'] = False

output_video_file = f'output_live.flv'

  

def video_frame_callback(frame: av.VideoFrame):
    frame = frame.to_ndarray(format="rgb24")  # Decode and get RGB frame
    frame, _ = live_process_frame.process(frame, pose)  # Process frame
    return av.VideoFrame.from_ndarray(frame, format="rgb24")  # Encode and return BGR frame


def out_recorder_factory() -> MediaRecorder:
        return MediaRecorder(output_video_file)


ctx = webrtc_streamer(
                        key="Squats-pose-analysis",
                        video_frame_callback=video_frame_callback,
                        rtc_configuration={"iceServers": [{"urls": ["stun:stun.l.google.com:19302"]}]},  # Add this config
                        media_stream_constraints={"video": {"width": {'min':480, 'ideal':480}}, "audio": False},
                        video_html_attrs=VideoHTMLAttributes(autoPlay=True, controls=False, muted=False),
                        out_recorder_factory=None
                    )