Object Detection

Use the Object Detection labelling task tool on DS2.ai to facilitate manual labelling.

Start labeling Object Detection

1. Create Labeling Project

1. Click on the Dataset in the top navigation bar.

2. Click to select the check box on the left side of the dataset for labeling. (If the dataset is not uploaded, refer to Upload Data and upload the data.)

3. Click the Start Labeling button at the top of the data list.

4. Enter a Project Name and Project Description(optional).

5. Set the Data Category to Object Detection and click the Next button.

2. Create a Class

  1. Click the Class tab of the Labeling Project.

  2. On the Class tab, click the Add Class button at the bottom left of the Tab menu.

  3. Set the Class Name and Class Color and click the Enter button. (For more Information, see Labeling > Labeling Project > Class.)

3. Perform manual labeling

At least one class must be created to perform manual labeling.

  1. Click on the Dataset in the top navigation bar.

  2. Select the labeling project that you created in the previous step.

  3. On the Dashboard tab of the labeling project, click the Start Manual Labeling button at the top.

Supported Labeling Format

DS2.ai supports bounding box, polyline, polygon, and magic tool labeling.

Bounding Box

Label in a rectangular box. Drag the mouse to draw a square area after you have hit the starting point.

Poly line

Labeling in the form of a line where the start and end points are not connected. Both linear and non-linear are available.


It refers to polygonal labeling that connects the start and end points. After taking a start point, return to the start point by marking each vertex.

Introduction to Object Identification Labeling Tool Features

Labeling Tool

Labeling Project Information.

  • Project Details Detailed description of the project and the progress of labeling are displayed.

  • Labeling Information After the project is created, you can check the specified labeling class.

  • Undo/Redo Cancels or reruns the most recent job history.

  • Shortcuts Displays a list of shortcuts currently set for each tool.

  • Ticklines When labeling, display or hide horizontal and vertical tick lines above the image file.

  • Labeling I have done You can check the number of labels for each of the bounding boxes, polygons, and magic tools that have been completed.

  • Tool-specific shortcuts are supported for easier labeling processes.

Do you have any other questions? [email protected]

Last updated