ArUco Markers

ArUco markers are binary square fiducial markers used for camera-based pose estimation.

Installation

pip install opencv-contrib-python

Detect Markers

import cv2
import cv2.aruco as aruco
import numpy as np

cap = cv2.VideoCapture(0)
aruco_dict = aruco.getPredefinedDictionary(aruco.DICT_4X4_50)
params = aruco.DetectorParameters()
detector = aruco.ArucoDetector(aruco_dict, params)

while True:
    ret, frame = cap.read()
    corners, ids, _ = detector.detectMarkers(frame)
    if ids is not None:
        aruco.drawDetectedMarkers(frame, corners, ids)
        print("Detected IDs:", ids.flatten())
    cv2.imshow("ArUco", frame)
    if cv2.waitKey(1) == ord('q'):
        break

Pose Estimation

# Camera calibration matrices (from calibration step)
camera_matrix = np.array([[fx, 0, cx],[0, fy, cy],[0, 0, 1]])
dist_coeffs = np.zeros((5,1))
marker_size = 0.05  # metres

rvecs, tvecs, _ = aruco.estimatePoseSingleMarkers(
    corners, marker_size, camera_matrix, dist_coeffs)

TIP

Print markers at exactly the size you specify as marker_size. Physical size accuracy directly affects translation accuracy.