This project demonstrates a simple hand-tracking and finger-counting application using Python, OpenCV, and Google's Mediapipe library. By utilizing real-time webcam input, the program detects the user's hand, identifies individual fingers, and accurately counts the number of fingers held up. The combination of OpenCV for video capture and image processing with Mediapipe’s efficient hand-tracking model allows for an effective, real-time finger-counting solution. This project serves as a foundational exercise in computer vision and hand gesture recognition.
import cv2
import mediapipe as mp
# Initialize MediaPipe hands and drawing modules
mp_hands = mp.solutions.hands
mp_drawing = mp.solutions.drawing_utils
# Setup camera capture
cap = cv2.VideoCapture(0) # Use 0 for the default camera
# Increase frame size (width=1280, height=720 for HD)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, 1280)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 720)
# Use the hands model for detecting hand landmarks
with mp_hands.Hands(max_num_hands=1, min_detection_confidence=0.7) as hands:
while cap.isOpened():
ret, frame = cap.read() # Read a frame from the camera
if not ret:
break
# Flip the frame horizontally for a mirror effect
frame = cv2.flip(frame, 1)
# Convert the BGR image to RGB
rgb_frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
# Process the image and detect hands
result = hands.process(rgb_frame)
# If hands are detected
if result.multi_hand_landmarks:
for hand_landmarks in result.multi_hand_landmarks:
# Draw landmarks on the hand
mp_drawing.draw_landmarks(frame, hand_landmarks, mp_hands.HAND_CONNECTIONS)
# Get landmark coordinates (wrist and tips of each finger)
landmarks = hand_landmarks.landmark
# Finger landmarks: [4, 8, 12, 16, 20] (Thumb, Index, Middle, Ring, Pinky)
finger_tips = [4, 8, 12, 16, 20]
fingers = []
# Check if fingers are open (landmark y-position of finger tip < that of knuckle)
for i in range(1, 5): # For fingers (excluding thumb)
if landmarks[finger_tips[i]].y < landmarks[finger_tips[i] - 2].y:
fingers.append(1) # Finger is open
else:
fingers.append(0) # Finger is closed
# Check thumb (landmark x-position for thumb)
if landmarks[finger_tips[0]].x < landmarks[finger_tips[0] - 1].x:
fingers.append(1) # Thumb is open
else:
fingers.append(0) # Thumb is closed
# Count the number of fingers that are open
finger_count = fingers.count(1)
# Display the number of fingers in orange color (BGR = (0, 165, 255))
cv2.putText(frame, f'Fingers: {finger_count}', (50, 50), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 165, 255), 2)
# Show the frame with the finger count
cv2.imshow("Finger Detection", frame)
# Break the loop if 'q' is pressed
if cv2.waitKey(1) & 0xFF == ord('q'):
break
# Release the camera and close all OpenCV windows
cap.release()
cv2.destroyAllWindows()
