Great news for people who can't find Waldo: At last, there's a robot with a silicone hand that can find him for us.
Creative agency Redpepper has built a robot that finds Waldo with tremendous accuracy. The robotic arm (uArm Swift Pro) is controlled by a Raspberry PI equipped with a Vision Camera Kit for facial recognition. The camera takes a photo of the page and then uses OpenCV to find Waldo's face in the photo. The photos are then analyzed by Google's AutoML vision service, which has been trained with images of Waldo beforehand. Once the calculated match is 95% or more, the arm pulls on Waldo.
The Google Cloud AutoML software has been available since January and allows users to train artificial intelligences without programming knowledge. With the drag'n'drop tool, anyone can craft an image recognition tool.
The whole thing usually runs in less than 5 seconds.