Skip to content

Auto-Labeling in General AI

1. What is Auto-Labeling in General AI?

General AI can proceed with auto-labeling without manual labeling by utilizing artificial intelligence from already developed labeling AI.

이미지이름

2. Detected objects and autolabeling types

1) Person Detection

Provides information by detecting the location of human objects in the image.

이미지이름

  • class : person
  • auto-labeling types: bounding box, polygon

2) Animal Detection

Detects and identifies specific objects of animals in the image and provides information.

이미지이름

  • class : person, bear, bird, cat, cow, dog, elephant, giraffe, horse, person, sheep, zebra (You can select and detect only the desired objects from the supported classes.)

  • auto-labeling types: bounding box, polygon

3) Self-driving

Detects and identifies specific objects in transportation and road facilities in the image and provides information.

이미지이름

이미지이름

  • class: person, airplane, bench, bicycle, boat, bus, car, firehydrant, motorcycle, parkingmeter, stopsign, trafficlight, train, truck (You can select and detect only the desired objects from the supported classes.)

  • auto-labeling types: bounding box, polygon, semantic segmentation

4) Face Point Detection

Display the face landmark recognizing one person or multiple people within the image.

이미지이름
Features such as nose, eyes, mouth, eyebrows, and jawline on a person's face can be called landmarks. Given the location and size of the face, it automatically determines the appearance of facial components, such as the eyes and nose, and provides results for application in a variety of ways.

  • class : person

  • auto-labeling types: bounding box, polygon

5) Keypoint Detection

Point coordinates to the joint of the person in the image and connect it with a line to detect key points.

이미지이름

Form a set of coordinates that can be linked to describe a person's pose. As shown in the picture above, it is possible to extract the person's skeleton by holding each keypoint with a person's joint joint and connecting the keypoint in the image or image.

  • class : person

  • auto-labeling types: bounding box, polygon

3. Create General AI model

[NOTE] Use the animal detection of General AI

3-1. Create a Project

이미지이름

1) Click the Start labeling button under "LABELING AI | Labeling" or go to "DS2 DATASET | Dataset".

2) Select the data set in "ZIP" format to auto-labeled and click the Start Labeling button. At this point, the upload status of the data must be "Complete" before the project can be generated normally.
Please check Upload Data for uploading data.

3) Select the project name, description, and data category as "object recognition," then click Next.

4) You can see that the project you created has been added to the project list under "LABELING AI | Labeling".

3-2. General AI Auto Labeling

이미지이름

Click the Start Autolabeling button on the Labeling Project Dashboard. Set the following options for General AI development.

1) Select artificial intelligence type

Please select the labeling artificial intelligence that you want to develop among people's keypoints such as Siram, animal, autonomous driving, face point detection, and human keypoints in General AI.

2) Select class

You can select which objects in each category you want to recognize. You must select at least one class and can select multiple classes.

3) Select auto labeling type

  • General (Box): Label objects in the form of rectangular boxes.
  • Polygun: Label objects linearly by dotting object attention according to the object's
  • Semantic: Label objects separated by pixels.

( Sophisticated labeling is possible in the order of General → Polygun → Semantic.)

4) Pre-processing Options

When you check the Face De-identification option, it detects all human objects in the image and mosaic the face to de-identify them.

5) Auto-labelling Longevity

You can autolabel the maximum number of uploaded images from a minimum of 100. After setting options, click the Start Autolabeling button to start autolabeling automatically.

3-3. Check Autolabeling Results

You can check the result yourself and modify the label.

이미지이름

Check "labeling ai | labeling → data list" for data whose job status is "Complete".

Click on the data you want to see in the list to see the auto-labeled work, and click [EDIT LABEL] at the bottom left to modify the label.