|▲ Datamaker / CEO Lee Enock|
(PowerKorea) The Ministry of Science and ICT recently held a briefing session on its 7 supplementary budget projects of 2020 that included AI, data and cloud. The ministry briefed that they had poured 1.6 trillion won (1.4 billion USD) in 81 projects and would pour further 44.8 trillion won until 2025. Among the 81 projects, AI (Artificial Intelligence) received the subsidy the most (292.5 billion won). The ministry put up public challenges for 150 AI tasks last year and the excellence went to Daejeon Metropolitan Express Transit Corporation and Datamaker. Datamaker proved its excellence in AI data labelling by installing a CCTV-based abnormal behavior detection system in Daejeon City Hall station.
Datamaker is a South Korean AI data learning and labelling company established in October 2018. It has made strong presence in AI industry by running data labs in South Korea and Ghana. According to experts at Datamaker, the level of perfection in AI learning hugely depends on both quantity and quality of data. In other words, learning AI requires data which should be rich in number and quality because the sorting out is then made into from simple calculation to complicated.
How do we make data for learning? Datamaker experts say it is like attaching eye balls to a digital doll by hands. Man should check every single data whether the image, for example, in the photo is a cat or a dog and where it is located; this process is called ‘labelling’. And this is what Datamaker is specialized at. Labelling being done, it is time for AI to learn and analyze.
Datamaker runs two data labs in Ghana including Accra, the capital city. Ghana’s minimum wage is South Korea’s 1/20. This means that the country can provide cost effective manpower which is the strongest competitiveness of companies like Datamaker. Datamaker used this data to build the abnormal behavior detection system collaborated with Daejeon Metropolitan Express Transit Corporation; the company labelled enormous amount of video images in the subway to build the system that detects and alarms abnormal behaviors such as violence and stealing. Daejeon City Hall station is the first to run this system in the nation.
AI data labelling in South Korea is in cooperative relationship with crowdsourcing as it requires large number of manpower who can carry out labelling even at home. And the labelled data in South Korea and Ghana is crisscrossed to create perfect AI data pools.
As of July 2020, Datamaker carried on labelling worth 5 billion won (4.4 million USD) required by public institutions. August last year, Datamaker participated in an AI data learning project worth 40 billion won, and in November the same year achieved 500,000 times of data processing a day by 10,000 labelers. Datamaker has paid 1.5 billion won to labelers as of 2021 and received a citation from Science and ICT Minister in recognition alongside being designated for 2021 Date Voucher Support Project. Companies request AI data learning to Datamaker is given 10 million won government subsidy.
Taking the momentum, Datamaker is introducing cutting edge labelling techniques: 3D point cloud frame, 3D LiDAR point cloud and 3D polyline as well as vided annotation and semantic segmentation.
“America is the biggest AI market in the world and the labelling is done in India for two reasons: cheap labor and English as an official language. My parents and I have carried on missionary works in Ghana for more than 20 years and I have extensive knowledge and experience in the country. 70 experts are working at the data labs in Ghana at the moment with thousands of resident labelers. I will keep use this excellent labor infra to grow presence of Datamaker in the field by turning Ghana as a typhoon of AI data labelling” says Lee Enock, CEO of Datamaker.
PowerKorea is an official partner of Herald Corporation. The magazine is distributed to the KOTRA business centers in the world. PowerKorea publishes latest news of government, Samsung, Hyundai and LG and endeavors to serve a platform to engage with people around the world.
신태섭 기자 firstname.lastname@example.org