What is data labeling?
1. What is data labeling?
In order to use artificial intelligence, various data must be injected into artificial intelligence to learn. At this time, artificial intelligence needs to classify and process data in a form that can be learned by itself, which is called data labeling.
Data labeling is a process that allows artificial intelligence to judge and learn data properly. Improve data quality by accurately labeling large amounts of data. If labelling is done incorrectly or learned with only a fraction of the data, the performance of artificial intelligence can be reduced. For example, what if an autonomous car recognizes people as street trees? Or we can think of a case in which artificial intelligence analyzing criminals recognizes a particular race as a criminal. Artificial intelligence services learned with these incorrect data are an important step to take care of because they can lead to dangerous and discriminatory factors.
2. Data Labeling at Labeling AI
Labeling AI supports labelling operations using various methods. You can choose how to label structured and unstructured data, such as images, speech, and text, for each situation and purpose. For image labeling, data labeling can be carried out using more advanced labeling functions such as simple and easy-to-use bounding boxes or polygons, skeleton, keypoints, semantic segmentation, and magic tools. Let's take a closer look at the differences in labeling methods supported by Labeling AI and the application criteria.