Data Labeling

Overview

The data labeling feature is deeply integrated with Label Studio — a powerful and flexible open-source data annotation tool. Through deep integration with the platform's dataset management module, all data import, management, and export is handled within the platform, providing unified data flow and a one-stop labeling experience.

Key Advantages

• Ready to Use: No separate Label Studio installation needed — open it directly within the platform
• Unified Data Management: All data import and export is handled through the platform's dataset management, ensuring consistency and traceability
• Multi-modal Support: Supports labeling of text, images, audio, video, HTML, and multi-sensor data
• Multi-format Export: Annotation results can be exported in multiple formats for model training or sharing
• Flexible Configuration: Supports Label Studio built-in annotation templates as well as custom labels and interfaces

Workflow

Step 1: Open the Labeling Tool

In Data Tools → Data Labeling, click to open the labeling tool. The system calls the backend interface (/dataflow/studio/jump-to-studio) to generate an access link and opens the Label Studio workspace in a new tab.

Step 2: Create a Labeling Project

In Label Studio, create a new project:

1. Enter a project name
2. Save the project to begin the new labeling task

Step 3: Import Data

Import data for labeling from your platform dataset:

1. In the Label Studio project, select Import Data
2. Choose the data branch from the platform dataset and import
3. Wait for data loading to complete

Step 4: Configure the Labeling Interface

After import, set up the labeling configuration:

• Use built-in templates: Label Studio provides templates for text classification, NER, image classification, object detection, and more for quick setup
• Custom labels: Define custom label types and annotation interfaces based on business requirements

Step 5: Perform Annotation

Once configured, begin labeling data item by item. Supported workflows:

• Single-annotator labeling
• Multi-person collaborative annotation (distribute tasks via project member management)
• Model-assisted pre-labeling (use model inference results to improve efficiency)

Step 6: Export Results

After labeling is complete, export results and save to the platform dataset:

1. Select Export in Label Studio
2. Choose an export format (JSON, CSV, etc.)
3. Results are automatically saved to the platform dataset with the branch suffix _label

Supported Annotation Types