Traditional tools for labeling text in images require that you manually adjust the box to fit the text. This is characterized by low efficiency, and can easily bring about labeling errors when the text is overly dense. Smart labeling of Machine Learning Platform for AI supports text box fitting, which can be used to automatically adjust skewed, distorted, and congested text boxes. Smart labeling also supports character recognition, which can automatically recognize content in a selected area and then convert it to text. This enables you to complete labeling jobs efficiently.
Smart text box fitting
- Manual labeling
Manual labeling is marked by low efficiency and cannot control the size of text boxes.
- Smart labeling
Text boxes can be scaled to fit the text by using the smart labeling feature. This allows you to improve the efficiency of data labeling.
Smart text recognition
- Manual labeling
After you select the content, you must enter the text, which leads to low efficiency.
- Smart labeling
Select the text content, and then click Smart Recognition. The system can automatically recognize the text in the box and generate corresponding characters.
Skewed text labeling
- Manual labeling
Polygon selection tools are used for labeling, which can be inefficient.
- Smart labeling
You only need to use the rectangle selection tool to select the skewed text area. The system can automatically identify the text range and fit the text.
Dense text labeling
- Manual labeling
When manual labeling is used in dense text, other text around the target text can be selected in the box, which may cause labeling errors.
- Smart labeling
To perform dense text labeling, you only need to use the rectangle selection tool to select the area that contains the dense text. The system can automatically fit the target text, even if undesired text is selected by mistake.
