All Products
Search

This Product

    Platform For AI:iTAG overview

    Document Center

    Platform For AI:iTAG overview

    Last Updated:Jun 21, 2026

    iTAG is PAI's data labeling platform. It offers templates for labeling various data types, including images, text, video, and audio, as well as for multimodal tasks.

    Supported labeling tasks

    iTAG provides built-in templates for the following labeling tasks:

    • Image: image classification, object detection, image OCR, table recognition, and image semantic segmentation.

    • Text: text classification, named entity recognition, and entity relationship recognition.

    • Video: video classification, video frame tagging, and video OCR.

    • Audio: audio classification, audio segmentation, and audio recognition.

    • Large model: visual question answering, multimodal RLHF annotation, image-to-text, image-text explanation, dialogue rewriting, dialogue sorting, and dialogue grouping.

    To use templates other than those built into the console, such as for text and image classification, see template management.

    Workflow

    1. Create a dataset

      Upload the data you need to label to Object Storage Service (OSS). Then, use the dataset management module to import the data and create a dataset. The system generates a .manifest index file—a JSONL file containing data paths and metadata—for labeling tasks.

      Important

      Currently, iTAG only supports data stored in OSS. To ensure proper access, the OSS bucket must be in the same region as PAI.

    2. Create a labeling task

      For an existing dataset, create and distribute a labeling task with a built-in or custom template. The task workflow consists of three stages: labeling, quality inspection, and acceptance. Labeling is required, while quality inspection and acceptance are optional. Each stage serves the following purpose:

      • Labeling: On the Label Task page, annotators claim a task package, complete the labeling, and submit it.

      • Quality inspection: On the Quality Inspection Task page, reviewers claim a completed task package to review, modify, or reject.

      • Acceptance: On the Acceptance Task page, project owners claim a task package for final review, modification, or rejection.

    3. Process the labeling task

      Follow the workflow to label, inspect, or accept task packages to produce labeled data.

    4. Export labeling results

      Export the labeling results to a specified OSS directory for model training. The results are exported in .manifest format.

    Billing

    • iTAG platform (Free): The iTAG platform is free of charge if you use your own team for manual labeling.

    • Intelligent labeling service (Free): The intelligent labeling service, which is available for certain large model templates like image-to-text and image-text explanation, is currently free. You will be notified in advance if this service becomes a paid feature.

    • Object Storage Service (OSS) (Paid): iTAG depends on Alibaba Cloud Object Storage Service (OSS). Therefore, costs for OSS storage and data read/write traffic are billed separately according to the OSS billing standards.

    • Manual labeling outsourcing service (Paid): To delegate data labeling to Alibaba Cloud's professional team, submit a ticket or join DingTalk group 21930006619 to contact the PAI team for this paid service.

    Get help

    If you encounter issues such as data loading errors, insufficient permissions, or need to configure cross-origin resource sharing (CORS) rules for OSS, see the iTAG FAQ.