Skip to main content

Data Annotation

Data Viewing​

Go to the Data page and click the corresponding file name link to start viewing and annotating data. The data page displays all available datasets, and you can quickly find the data you need using the search and filter functions.

The dataset page displays all available data files

Panel​

A panel is the rendering unit of data, similar to a camera perspective (Topic).

Each data entry usually contains multiple camera perspectives, which are represented by multiple panels. You can flexibly adjust the size, position, and number of panels to browse the data in the optimal way. The flexible layout allows you to view data from different angles simultaneously, significantly improving annotation efficiency.

Switching Render Panels​

You can easily change the content displayed in any panel:

  1. Click the settings button in the upper left corner of the panel
  2. Select the desired panel type from the dropdown menu

This allows you to quickly switch between different data views according to your current task needs.

Click the settings button in the upper left corner of the panel

Select the desired panel type from the dropdown menu

Changing Subscription Topics​

You can also change the data source displayed in the panel:

  1. Click the settings button in the upper left corner of the panel
  2. In the pop-up settings menu, select the Topic message source you want to switch to

This enables you to view different data streams to meet various annotation scenarios.

Change the Topic message source through the settings menu

Preview the newly selected Topic content

Fine-grained Annotation​

tip

The annotation method in this section is suitable for segment annotation containing multiple atomic actions, especially when a single data entry contains multiple tasks. Fine-grained annotation allows you to mark multiple different actions or events within a single data stream.

For annotation task workflow, please refer to:

Start Annotation​

Click the "Start" button (or press shortcut key Q) to mark the start time point of an action or event. The system will record the current timestamp as the start point of the annotation.

Click the start button or use shortcut Q to start annotation

You can:

  • Play automatically (press Space)
  • Manually jump forward/backward (press Left/Right arrow keys)
  • Control jump speed (hold CTRL to speed up or ALT to slow down)

These flexible playback controls help you precisely locate the time point to annotate.

End Annotation​

When the action or event ends, click the "End" button (or press shortcut key R) to end the current annotation. The system will record the current timestamp as the end point, thus determining the complete time range.

Click the end button or use shortcut R to end annotation

Add Semantic Information​

After marking the time segment, you need to add semantic information for the annotation:

  1. Fill in the form information
  2. Select the corresponding natural language description or add a new dictionary entry
  3. Save the information

Detailed semantic descriptions help with subsequent data training and analysis.

Fill in the natural language description for the annotation Select a predefined description or add a new one

Add Annotation​

After completing the above steps, click the "Add Annotation" button (or press Enter) to add this annotation to the current data's annotation list. If you need to annotate multiple actions, repeat the above steps.

Click the add annotation button or use Enter to confirm

Save Annotation​

After completing all annotations, click the yellow upload button in the upper right corner to save the annotations to the server. Please make sure to save your work before leaving the page to prevent data loss.

Click the save button in the upper right corner to upload annotations to the server

Batch Annotation​

tip

This feature allows you to apply the annotation results of one data entry to multiple similar homogeneous data entries, significantly saving annotation time.

For single-task scenarios (such as 100 data entries with the same action), the batch annotation feature can greatly improve efficiency.

Batch Annotation Process​

  1. Select Sample Data: Batch select homogeneous data from the data list Select a data entry as a sample from the list

  2. Create Annotation Task: Create a new annotation task for these data entries Create a new annotation task

  3. Start Annotation: View and start executing the annotation task View the created annotation task

  4. Complete Sample Annotation: Complete the annotation of a sample according to the fine-grained annotation process Complete the annotation of the sample data

  5. Save Sample: Save the completed sample annotation Save the completed sample annotation

  6. Select Template: Return to the task page and select the annotated sample data as a template Select the sample to use as a template

  7. Batch Copy: Start batch copying the annotation to other data entries Start the batch copy annotation process

  8. Confirm Success: Confirm that the copy operation was successful System shows annotation copy succeeded

  9. View Results: View the final results of the batch annotation View the final results of the batch annotation

With the batch annotation feature, you can quickly apply standardized annotations to large amounts of data, which is especially suitable for handling large-scale homogeneous datasets and greatly improves work efficiency.