webcam_gesture_recognition.py

import cv2
import mediapipe as mp
import numpy as np
import joblib

# Load your trained model and label encoder
model = joblib.load("gesture_model.pkl")
label_encoder = joblib.load("label_encoder.pkl")

# Initialize MediaPipe Hands
mp_hands = mp.solutions.hands
mp_drawing = mp.solutions.drawing_utils

hands = mp_hands.Hands(
    static_image_mode=False,
    max_num_hands=1,
    min_detection_confidence=0.7,
    min_tracking_confidence=0.7)

cap = cv2.VideoCapture(0)  # 0 = default webcam

while cap.isOpened():
    ret, frame = cap.read()
    if not ret:
        print("Ignoring empty camera frame.")
        continue

    # Flip frame horizontally for selfie-view and convert BGR to RGB
    frame_rgb = cv2.cvtColor(cv2.flip(frame, 1), cv2.COLOR_BGR2RGB)

    # Process the frame to detect hands
    results = hands.process(frame_rgb)

    # Convert back to BGR for OpenCV
    frame = cv2.cvtColor(frame_rgb, cv2.COLOR_RGB2BGR)

    if results.multi_hand_landmarks:
        for hand_landmarks in results.multi_hand_landmarks:
            # Draw hand landmarks on frame
            mp_drawing.draw_landmarks(frame, hand_landmarks, mp_hands.HAND_CONNECTIONS)

            # Extract landmarks as a flat list [x0, y0, z0, x1, y1, z1, ...]
            landmarks = []
            for lm in hand_landmarks.landmark:
                landmarks.extend([lm.x, lm.y, lm.z])

            # Convert landmarks to numpy array for prediction
            landmarks_np = np.array(landmarks).reshape(1, -1)

            # Predict gesture label using your model
            prediction = model.predict(landmarks_np)
            predicted_label = label_encoder.inverse_transform(prediction)[0]

            # Display the predicted label on the frame
            cv2.putText(frame, f'Prediction: {predicted_label}', (10, 30),
                        cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)

    # Show the frame
    cv2.imshow('Gesture Recognition', frame)

    # Exit loop on 'ESC' key press
    if cv2.waitKey(5) & 0xFF == 27:
        break

# Release resources
cap.release()
cv2.destroyAllWindows()
hands.close()