Manfu Technology: seven commonly used data tagging tools

If you want to do good work, you must first sharpen its tools.

Labeling tools are the foundation of the data labeling industry, and a useful labeling tool is the key to improve labeling efficiency and produce high-quality tagged data.

The commonly used data annotation tools are as follows: 2D box, semantic segmentation, polygon segmentation, point annotation, line annotation, video annotation, 3D cube annotation and so on.

1. 2D box

The 2D box is rectangular, and of all the annotation tools, the 2D box is the simplest type of data annotation and the lowest cost.

two。 Semantic segmentation

Speech segmentation is not only a more accurate annotation type in the field of image annotation, but also a time-consuming annotation type. The annotator needs to label all the content on the picture.

3. Polygon segmentation

Compared with 2D frame, polygon segmentation is used for accurate object detection and location in pictures and videos. Compared with 2D boxes, polygon segmentation is more accurate, but it is also more time-consuming and more expensive.

4. Point callout

Point labeling is to determine the shape changes of large and small objects through multiple continuous points, which is usually used in statistical models and posture or facial recognition models.

5. Line callout

Line marking is mainly used for road identification of self-driving vehicles, defining different roads such as vehicles, bicycles, traffic lights in the opposite direction, forks and so on.

6. Video annotation

Video annotation, that is, locating and tracking objects in a series of images in terms of frames. It is mainly used to train self-driving prediction models such as vehicles, pedestrians, cyclists, roads and so on.

7. 3D cube dimension

3D cube annotation is used to obtain spatial visual model from 2D pictures and videos, measure the relative distance between objects and get vanishing points.