PINGDINGSHAN, China – (Reuters) – In a small village in central China’s Henan province, surrounded by barking dogs and wandering chickens, the villagers gather along a dirt road to the trading of images of their faces, for use with boilers of pots, and of cups of tea.
FILE PHOTO: Employees work on the identification of the various items required for the collection of data on computer screens that would be used for the development of an artificial intelligence (AI) and machine learning technology, and the Ji Qian of the Data, Co is in Jia county, Henan province, China, March 20, 2019 at the latest. REUTERS/by Irene Wang
On the front of the line, a woman is standing in front of a camera zip-tied to a tripod. It has a picture of the head, the eyes, the nose and the cut out in the side of her face, and slowly turn from left to right.
The villagers are waiting for their turn to have a numbered ticket. Some of them say that it was the third or fourth time that they have come to do this kind of work.
The project, carried out by a sleepy courtyard, village house, decorated with posters of the former China’s leader, Mao Zedong, the development of materials that would be able to train with AI software, in order to distinguish between the facial features, and still images.
“The major projects, tens of thousands of people that live in this area,” said Liu Yangfeng, the chief executive of the Qianji Information Co., Ltd, which collects and tags the data with some of China’s biggest tech companies, and is located in the city of Pingdingshan.
“We are providing more data sets and the need for more AI algorithm, and the companies so that they can be used for the development of artificial intelligence in China,” said Liu, declining to disclose any of its customers.
The increase in the demand for data in order to train the AI algorithms, as is the feeding of a new world-wide industry in which it collects information, including photos and videos, which are then labeled in order to tell the machines what they are seeing.
The companies that are active in the data label or data annotation, as it is also called, are crowdsourcing platforms, such as Amazon.com’s (AMZN.(O) to the Mechanical Turk users for a small amount to you in exchange for simple tasks, outsourcing companies such as india’s Wipro Ltd (WIPR.NS), and professional, and come as is Qianji.
Cognilytica, a US-based research firm specializing in AI) estimates that the global market for machine-learning-related-data, annotation, increased by 66% to $500 million by 2018 and to more than double by 2023. Some industry insiders say, however, that much of the work is done, it is not known, making accurate estimates difficult.
WEAK PROTECTION LAWS, CHEAP LABOR
China has emerged as an important hub for data collection and labeling, thanks to the insatiable demand of the emerging artificial-intelligence sector, backed by the ruling Communist y, which sees AI as an engine of economic growth and as an instrument for social control.
A lot of companies have invested in the area of AI known as machine learning, which is at the core of face recognition technology and systems, which are based on finding patterns in the data.
These include tech giants Alibaba Group Holding Ltd (BABA).(N), Tencent Holding Ltd’s (0700.HK), Baidu Inc (BIDU.(O), as well as younger companies such as AI specialist will SenseTime Group, inc., and voice-recognition firm Iflytek Co Ltd (002230.SERIES).
The result has been the proliferation of COMPUTER-related products and services in China, a face recognition-based-payment-systems-for-monitoring-and even COMPUTER-animated, state-media news, posts. The Chinese consumers you will typically see these technologies as a new and futuristic, in spite of the objections of some of the more invasive apps.
The weak data privacy laws and cheap labour have a competitive advantage for China as it races to become a global leader in the AI. The Henan villagers were very happy to see the market from a number of sessions with a video camera in the world with a cup of tea, or a few hours for the stove and the pot.
The Beijing-based BasicFinder, one of the leading data labeling as well as a company with operations in Hebei, Shandong and the provinces of Shanxi, has a robust mix of domestic and foreign customers.
On a recent visit to the Beijing offices, and some of the people have been labeling as well as images of the dreamy people, who are going to be used by the autonomous driving project is to identify drivers who may be falling asleep behind the wheel.
The others were the book of the British documents dating back to the 1800s, before the West online, lineage, service, marker, fields, dates, names, and families in the birth and death records.
According to the BasicFinder the Chief Executive of Du, Lin, hired, trained, and labeling machines in China, it is less expensive than the use of a Western crowd sourcing marketplace.
A Princeton University project in the field of autonomous driving in the first place a task on Amazon’s Mechanical Turk, but the task was becoming more complicated, people began to make mistakes, and BasicFinder was brought in to assist with the correction of the results, said Du.
In this project, is a trained BasicFinder labeler has been able to do the work of three crowdsourced and labeling, he added.
“Gradually, they saw that they had less and less to pay for the labeling, so they hired us to label all the work, right from the start,” said Du.
The airport declined to comment.
For the identification of the employee, the reasons for their membership of the chinese economy are apparent. In the work, but sometimes it’s annoying, but it is an upgrade of the other jobs available to younger workers who want to return to the small size of Chinese cities, towns and villages in the region.
And come on Qianji about 100 yuan ($14.50) a day to mark the data points on the photos of the people, for the surveillance footage and street footage.
The work is usually very simple, according to the workers, even though some of the overseas content is a challenge.
Slideshow (4 Images)
“In a time when we thought we’d lay out a Europe-style cooker equipment with a ring attached,” said Jia Yahui, a labeler to Qianji. “Later on, we were told that it was in fact two separate things, a cooker and a washing machine.”
The identification of the work of a number of of the benefits of the tech sector in the rural areas, but these benefits may prove short-lived as the AI will improve enough to perform many of the tasks, and come to do.
“We believe that the industry will continue to exist in the three-to five-year period. It may not be a long-term, we can only imagine that the five-year plan, for now,” he said Qianji CEO in September.
Report by Cate Cadell; Editing by Jonathan Weber and Edwina Gibbs