What are the categories of data labeling?

Data annotation includes: classification annotation, target detection annotation, instance segmentation annotation, key point annotation, relationship annotation, etc.

1, classification and labeling

Classification annotation is one of the most common types of data annotation, which classifies data into different categories according to their characteristics. For example, in image recognition, classification labeling can divide images into different categories, such as cats, dogs and people. In natural language processing, text can be classified into positive, negative or neutral emotional categories by classification tagging.

2. Target detection and marking

Target detection labeling refers to labeling a specific target object in an image or video and giving its position, shape and other information. This marking method is often used in automatic driving, monitoring system and other applications.

3. Instance segmentation and labeling

The annotation of instance segmentation refers to the segmentation of different instances in the image, and gives the bounding box and label of each instance. This marking method is often used to detect people or objects in the fields of autonomous driving, robot vision and so on.

4. Key point marking

Key point labeling refers to labeling specific key points in images or videos, such as faces and body parts. This labeling method is often used in face recognition, behavior analysis and other fields.

5. Relationship label

Relationship labeling refers to the relationship between different entities, such as the relationship between people and things, the relationship between things and so on. This labeling method is often used in knowledge mapping, natural language processing and other fields.