Camera board • Aruco pose estimation using PIcamera2 and Camera module 3

Code:

import cv2from cv2 import arucoimport numpy as npimport msgpack as mpimport msgpack_numpy as mpnARUCO_PARAMETERS = aruco.DetectorParameters()ARUCO_DICT = aruco.getPredefinedDictionary(aruco.DICT_ARUCO_ORIGINAL)detector = aruco.ArucoDetector(ARUCO_DICT, ARUCO_PARAMETERS)markerLength = 0.05markerSeperation = 0.01board = aruco.GridBoard(        size= [1,1],        markerLength=markerLength,          markerSeparation=markerSeperation,        dictionary=ARUCO_DICT)_video_pth = 'video path'_video_file = open(_video_pth, "rb")_video_data = mp.Unpacker(_video_file, object_hook=mpn.decode)_video_length = 0for _frame in _video_data:    _video_length += 1_video_file.close()# Selecting random 150 framesmarker_corners = []marker_ids = []counter = 0rnd = np.random.choice(_video_length, 150, replace=False)for idx, _frame in enumerate(_video_data):        if idx in rnd:        corners, ids, rejected_image_points = detector.detectMarkers(_frame)        corners, ids, _, _ = detector.refineDetectedMarkers(_frame, board, corners, ids, rejected_image_points)         if ids is not None:            marker_corners.append(corners)            marker_ids.append(ids)            counter+=1_video_file.close()# processing the cornersprocessed_image_points = []processed_object_points = []for _f in range(len(marker_corners)):    current_object_points, current_image_points = board.matchImagePoints(marker_corners[_f], marker_ids[_f])    try:        if current_object_points.any() and current_image_points.any():            processed_image_points.append(current_image_points)            processed_object_points.append(current_object_points)    except:        pass        mtx2 = np.zeros((3, 3))dist2 = np.zeros((1, 8))# camera calibration _, mtx1, dist1, _, _=cv2.calibrateCamera(processed_object_points, processed_image_points, _frame.shape[:2], mtx2, dist2)