Link Search Menu Expand Document

Introduction to AR Tags

Learn how to generate and detect AR Tags using Python and OpenCV.

Table of contents
  1. Requirements
  2. Code Examples and Data
  3. Generating AR Tags
    1. CHALLENGE 1: Generating AR Tags
  4. Detecting AR Tags
    1. CHALLENGE 2: Calculating Distance Between Tags


This practical requires the library opencv-contrib-python, which has additional modules that are not included in opencv-python. Install with the following command:

pip3 install opencv-contrib-python

Code Examples and Data

The data and examples used in this practical will be here. Clone this repository and use it as a boilerplate to complete the following challenges.

Generating AR Tags

You can generate and print your own AR tags for any use case. ArUco contains a number of AR Tag Libraires:

  • DICT_4X4_100
  • DICT_4X4_1000
  • DICT_4X4_250
  • DICT_4X4_50
  • DICT_5X5_100
  • DICT_5X5_1000
  • DICT_5X5_250
  • DICT_5X5_50
  • DICT_6X6_100
  • DICT_6X6_1000
  • DICT_6X6_250
  • DICT_6X6_50
  • DICT_7X7_100
  • DICT_7X7_1000
  • DICT_7X7_250
  • DICT_7X7_50

The tags of format DICT_{SIZE}x{SIZE}_COUNT are ArUco based tags that use SIZExSIZE bits for tag information and have COUNT distinct tags. We can use the ArUco library to create tags:

import numpy as np
import cv2
arucoDict = cv2.aruco.Dictionary_get(cv2.aruco.DICT_ARUCO_ORIGINAL)
SIZE = 500 # pixels
marker = np.zeros((SIZE, SIZE, 1), dtype=np.uint8)
ID = 0
cv2.aruco.drawMarker(arucoDict, ID, SIZE, marker, 1)
cv2.imwrite('DICT_ARUCO_ORIGINAL_id_{}_{}.png'.format(ID, SIZE), marker)

CHALLENGE 1: Generating AR Tags

Generate AR tags for the dictionary DICT_APRILTAG_16H5 for ids 7, 18, and 23.

  • These should be of size 500 pixels by 500 pixels.

Detecting AR Tags

The ArUco library has built-in functionality for detecting AR tags within images:

tags = cv2.imread('data/two_tags_ARUCO_ORIGINAL.png')
arucoDict = cv2.aruco.Dictionary_get(cv2.aruco.DICT_ARUCO_ORIGINAL)
corners, ids, rejects = cv2.aruco.detectMarkers(cv2.cvtColor(tags, cv2.COLOR_BGR2GRAY), arucoDict)
detection = cv2.aruco.drawDetectedMarkers(tags, corners, borderColor=(255, 0, 0))
cv2.imwrite('detection_two_tags_ARUCO_ORIGINAL.png', detection)

CHALLENGE 2: Calculating Distance Between Tags

Calculate the distance between the AR tags in the file data/two_tags_APRILTAG_16H5.png in centimeters. Some notes to keep in mind:

  • The real-world width of each tag is 3.
  • The dictionary being used is APRILTAG_16H5.
  • The image view is exactly perpendicular to the camera.
  • Both cv2 and numpy are useful packages here.

Back to top

© Copyright 2022 MIT Beaver Works Summer Institute.