Crowdworks (CEO Park Min-woo), an ArtificialIntelligence (AI) training data platform company, recently announced that ithad completed the registration of a US patent for 'Technology for AI targettraining image sampling’.
This technology is related to sampling the targettraining image from the video. It provides a method of sampling the target trainingframe image of the video for AI image learning that can minimize time and costin the pre-processing.
In case of using this patented technology, onlynecessary data can be selected when selecting learning data, and it is possibleto accurately measure the amount of change in the learning object by separatingthe target object and the background. This technology of Crowdworks wasrecognized for its excellence and was quickly registered through the PatentProsecution Highway (PPH) system.
Crowdworks is the largest patent holder inthe data labeling technology field in Korea, applying for over 130 domestic andforeign technology patents essential to the advancement of data labelingtechnology and securing market competitiveness.
A total of 50 domestic and foreign patentapplications and registrations were completed last year alone, including 18 newdomestic patent applications, 20 domestic registered patents, 11 new internationalpatent applications and 5 international registered patents. Crowdworks has 9Japanese patent registrations and 5 applications, such as the pre-processinginspection method and the Autonomous driving training image extraction method.
Starting with the first US patentregistration this year, the strategy is to continuously secure US patents andenter the North American market in earnest.
Crowdworks’ CEO Park Min-woo said, "Ascompetition in the domestic and global data labeling market intensifies, it isessential to secure technology-oriented competitiveness." And he continued; "We are going to lead the global AI training data market by activelysecuring intellectual property rights."