Manually identifying images to be used in the training phase of an AI system. Data labeling is an essential component in AI development as it identifies all the objects in the world people may be interested in when requesting information from a chatbot or creating new images and videos from a prompt.
All Around the World
It is estimated that more than a hundred million people worldwide do data labeling. It is performed in the U.S. as well as India, China, The Philippines and countries in Africa and South America where labor rates are low. Labeling is also done in Laos, Cambodia, Bangladesh and Bhutan.
Data labeling is a tedious job requiring people to sit at computers all day and type the names of things they see on screen. Labelers isolate each item in the scene with the cursor and assign an identifying name to it such as person, animal, car, truck, building and street sign. See
supervised learning.