Code Editor for app.py

import streamlit as st
import mediapipe as mp
import cv2
import numpy as np
from mediapipe.framework.formats import landmark_pb2

# Initialize MediaPipe Pose
mp_pose = mp.solutions.pose
mp_drawing = mp.solutions.drawing_utils
pose = mp_pose.Pose(
    min_detection_confidence=0.5,
    min_tracking_confidence=0.5,
    model_complexity=2  # Using the full model for better accuracy
)

# Function to calculate distance between two 3D points
def calculate_distance(point1, point2):
    return np.sqrt(
        (point1.x - point2.x)**2 + 
        (point1.y - point2.y)**2 + 
        (point1.z - point2.z)**2
    )

# Function to calculate real-world distances based on shoulder width reference
def get_scale_factor(landmarks):
    left_shoulder = landmarks.landmark[mp_pose.PoseLandmark.LEFT_SHOULDER]
    right_shoulder = landmarks.landmark[mp_pose.PoseLandmark.RIGHT_SHOULDER]
    shoulder_distance = calculate_distance(left_shoulder, right_shoulder)
    # Using 35cm as reference for shoulder width
    return 35.0 / shoulder_distance

# Define the connections we want to measure
CONNECTIONS = [
    (mp_pose.PoseLandmark.LEFT_SHOULDER, mp_pose.PoseLandmark.RIGHT_SHOULDER, "Shoulder width"),
    (mp_pose.PoseLandmark.LEFT_SHOULDER, mp_pose.PoseLandmark.LEFT_ELBOW, "Left upper arm"),
    (mp_pose.PoseLandmark.RIGHT_SHOULDER, mp_pose.PoseLandmark.RIGHT_ELBOW, "Right upper arm"),
    (mp_pose.PoseLandmark.LEFT_ELBOW, mp_pose.PoseLandmark.LEFT_WRIST, "Left forearm"),
    (mp_pose.PoseLandmark.RIGHT_ELBOW, mp_pose.PoseLandmark.RIGHT_WRIST, "Right forearm"),
    (mp_pose.PoseLandmark.LEFT_HIP, mp_pose.PoseLandmark.RIGHT_HIP, "Hip width"),
    (mp_pose.PoseLandmark.LEFT_HIP, mp_pose.PoseLandmark.LEFT_KNEE, "Left thigh"),
    (mp_pose.PoseLandmark.RIGHT_HIP, mp_pose.PoseLandmark.RIGHT_KNEE, "Right thigh"),
    (mp_pose.PoseLandmark.LEFT_KNEE, mp_pose.PoseLandmark.LEFT_ANKLE, "Left shin"),
    (mp_pose.PoseLandmark.RIGHT_KNEE, mp_pose.PoseLandmark.RIGHT_ANKLE, "Right shin"),
]

st.title("Real-time Pose Analysis with Measurements")
st.write("Stand in front of the camera to see your skeleton with real-time measurements!")

# Initialize the webcam
camera = st.camera_input("Pose Analysis Camera")

if camera is not None:
    # Convert the image from BGR to RGB
    bytes_data = camera.getvalue()
    cv2_img = cv2.imdecode(np.frombuffer(bytes_data, np.uint8), cv2.IMREAD_COLOR)
    image = cv2.cvtColor(cv2_img, cv2.COLOR_BGR2RGB)
    
    # Process the image
    results = pose.process(image)
    
    if results.pose_landmarks:
        # Get scale factor based on shoulder width
        scale_factor = get_scale_factor(results.pose_landmarks)
        
        # Create a copy of the image to draw on
        annotated_image = image.copy()
        
        # Draw the pose landmarks
        mp_drawing.draw_landmarks(
            annotated_image,
            results.pose_landmarks,
            mp_pose.POSE_CONNECTIONS,
            mp_drawing.DrawingSpec(color=(245,117,66), thickness=2, circle_radius=2),
            mp_drawing.DrawingSpec(color=(245,66,230), thickness=2, circle_radius=2)
        )
        
        # Draw measurements for each connection
        for connection in CONNECTIONS:
            start_point = results.pose_landmarks.landmark[connection[0]]
            end_point = results.pose_landmarks.landmark[connection[1]]
            
            # Calculate real-world distance
            distance = calculate_distance(start_point, end_point)
            real_distance = distance * scale_factor
            
            # Calculate the midpoint for text placement
            mid_x = int((start_point.x + end_point.x) * image.shape[1] / 2)
            mid_y = int((start_point.y + end_point.y) * image.shape[0] / 2)
            
            # Draw the measurement text
            cv2.putText(
                annotated_image,
                f"{real_distance:.1f}cm",
                (mid_x, mid_y),
                cv2.FONT_HERSHEY_SIMPLEX,
                0.5,
                (255, 255, 255),
                2
            )
        
        # Convert back to BGR for display
        annotated_image = cv2.cvtColor(annotated_image, cv2.COLOR_RGB2BGR)
        
        # Display the annotated image
        st.image(annotated_image, channels="BGR", use_column_width=True)
        
        # Display measurements in a table
        measurements = []
        for connection in CONNECTIONS:
            start_point = results.pose_landmarks.landmark[connection[0]]
            end_point = results.pose_landmarks.landmark[connection[1]]
            distance = calculate_distance(start_point, end_point)
            real_distance = distance * scale_factor
            measurements.append([connection[2], f"{real_distance:.1f}cm"])
        
        st.table(measurements)
    else:
        st.warning("No pose detected. Please make sure you are visible in the camera.")