Trend
"Data labeling" refers to the task of processing data into a form that AI can learn on its own to improve its algorithms. The people who help with this are called data labelers.
Data labelers label and attribute things like photos, animals in videos, objects, and more. As AI learns from this data, it recognizes similar images. There is a lot of work being done by "data labellers" to ensure that data is labeled with completeness.
It hasn't been long since data labelers emerged as a new profession, labeling a wide variety of data with names and attributes that AI can understand. The number of data labelers who have gone beyond the idea of a job or side hustle to a full-time career is growing rapidly.
According to the 2021'Data Labeler' Report, 46% of active 'data labelers' in Korea are making it their full-time job.
The expression and perception of "putting eyeballs on big data" can be interpreted as a reversal of the low evaluation of "data labeling". Here are some reasons why 'data labeler' is gaining traction.
First, it's a broad project, not a one-off task.
Second, the steady increase in demand from companies for "data labelers
Third, a fast-growing industry with an ever-increasing demand for data labelers.
Based on the top three companies that hire data labelers, the average hourly wage for entry-level data processing jobs is around $17,000, while the average hourly wage for advanced projects is as high as $25,000.
In addition, professional 'data labelers' who are qualified and familiar with the work can take home a large amount of money. And as the startup of data labeling is growing rapidly, it is attracting the attention of many people.
In Gwangju Metropolitan City, where interest in and demand for digital jobs is on the rise, 586 people completed data labeling training in the first year of its implementation in 2020, and 341 people worked as data labelers.
This was largely due to the increasing demand for data labeling from companies, which has resulted in 1,600 data processing workers in the past two years. Above all, the pre-processing work of labeling simple data such as text, voice, and images so that computers can learn on their own, and building data for learning, is easy for many people, including job seekers and career breakers, to get a job with simple training.
Based on the evaluation results, those who have completed the relevant training can participate in the data processing work of the '2022 Data Construction Support Project forAI Learning' of the Ministry of Science and ICT and the National Institute forIntelligence and Information Society (NIA) and receive digital jobs that allow them to work on-site or from home.
'Data labelers' who work on advanced projects will be recognized as skilled and receive specialized work. For example, data labelers with high TOEFL scores can work on English data labeling projects. In this case, they will be recognized as skilled and will receive a higher activity fee than the average 'data labeler'.
In addition, sports majors can participate in a slow-motion video collection project related to sports and do data labeling according to their experience. We conducted a survey on "Do you know about 'data labeling'?" among 500 general people living in Seoul aged 20-59, and 59% of them answered that they were aware of 'data labeling'.
The data labeling market is showing remarkable growth in the industry, and we expect the data labeling market to expand steadily. In response, we are working to revitalize the ecosystem by expanding data processing areas and industries.
At Datahunt, we are also thinking a lot about data that can be easily accessed by various people.In addition, we are making great efforts to provide data with high completeness.
For example, we are adopting the human-in-the-loop (HITL) method, in which human experts adjust intermediate results with the help of A.I., and A.I. checks them again, and we will continue to do a lot of research and development.