Where's Wally? Statistics can tell you!


  • Author: Carlos Grajales
  • Date: 11 October 2018
  • Copyright: Image appears courtesy of ClipArt Library

It is becoming increasingly common to hear of new ways statistics are changing our current landscape: from personal assistants driven by machine learning models to driverless cars that could potentially change the way we envision cities. With all the news containing more and more examples of stats breaking molds, it is not surprising to miss some seriously, life-changing developments in the field, such as the moment you discover stats are being used to find Wally [1].

thumbnail image: Where's Wally? Statistics can tell you!

If you missed your childhood, let me tell you who Wally is. Also known as Waldo in the U.S., Wally is the iconic character of a series of animated books created by British illustrator Martin Handford. The saga became so popular that the original books sold close to 20 million copies in their first four years in the market [2]. Unlike other children books, Handford’s titles contain hardly any text; they are basically a collection of two-page sketches filled with hundreds of characters in funny situations and somewhere, within all that mess, Wally is hiding, waiting eagerly for brave eyes to spot him. Now, modeling tools are being incorporated into a mechanic robot arm that can scan a book’s page and point with its own robo-fingers the exact location of Handford’s iconic character.

The process to train the robot is nothing remarkable, at least considering current Big Data standards. The authors googled over 60 Wally’s heads and other 45 Wallys’ full body images [1]. These images were used to train the model, using these as “success” cases in a binary classification problem. The model learned the common characteristics of Wally (quite a characteristic face in my opinion) and used those learned traits to compare it with the faces found in a full two-page image of a Wally’s book. When faced with such an exercise, the robot first takes a full-size photograph of the book and then uses the OpenCV library, originally developed by Intel, to identify all faces in the image [1]. Finally, the robot relies on the fitted machine learning model to distinguish Wally’s characteristics and separate them from the rest of the cartoon characters in the books. A metal robotic arm then moves to point out the exact location of Wally within the pages [1]. In the end, the model achieved quiet a remarkably accuracy. It took the developers about a week to code the robot in Python, yet in its fastest exercise the robot was able to point Wally in less than 5 seconds.

The system was designed and developed within a creative agency called Red Pepper [3]. Their tool is also representative of a new reality: the modeling process was developed using Google’s Cloud AutoML, a recently released Google tool that allows users to adapt some pre-trained image recognition models for specific tasks, as in this case, finding a quirky guy with glasses and red-stripes clothing [4]. Google’s AutoML cloud solution does not require programming expertise and most of its functions can be accessed with a simple user interface that allows users to drag and drop images on the tool to start training their own image recognition models. Besides image recognition tasks, in its Beta Phase Google AutoML allows users to experiment with pre-loaded models for Translation and natural Language Processing tasks. As you can see, these tools and the current democratization of analytical models are allowing new and original applications, some aiding the intelligent implementations of systems and some ruining our infancy as well.


[1] Lee, Dami. This robot uses AI to find Wally, thereby ruining Where’s Wally. The Verge Website (Aug, 2018)

[2] Stivers, Cyndi. Where’s Waldo? Entertainment Weekly Website (Dec, 1990).

[3] RedPepper - A Creative Agency built to Innovate. Redpepper Website

[4] Cloud AutoML BETA. Google Cloud Website

Related Topics

Related Publications

Related Content

Site Footer


This website is provided by John Wiley & Sons Limited, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ (Company No: 00641132, VAT No: 376766987)

Published features on StatisticsViews.com are checked for statistical accuracy by a panel from the European Network for Business and Industrial Statistics (ENBIS)   to whom Wiley and StatisticsViews.com express their gratitude. This panel are: Ron Kenett, David Steinberg, Shirley Coleman, Irena Ograjenšek, Fabrizio Ruggeri, Rainer Göb, Philippe Castagliola, Xavier Tort-Martorell, Bart De Ketelaere, Antonio Pievatolo, Martina Vandebroek, Lance Mitchell, Gilbert Saporta, Helmut Waldl and Stelios Psarakis.