Skip to main content
AI Workbench: Tag Log
Updated over 3 weeks ago

Overview

The AI Workbench: Tag Log module enables users to inspect video content by reviewing and modifying the tagging and classification applied by the AnyClip Platform during initial video processing. This tool provides a detailed breakdown of detected tags and allows manual adjustments to improve tagging accuracy.

Tag Log Interface

The interface consists of several key sections:

1. Video Frames and Tags Panel

  • Located on the right side of the screen, this panel displays a list of video frames that have been marked with system-generated tags.

  • The tags applied to each frame include:

    • System Tags: People, Brands, Brand Safety.

    • Keywords: AI-detected relevant words.

    • Text Tags: Extracted textual elements from the video.

  • These tags represent the objects, individuals, and textual content recognized by the AnyClip Platform’s AI services in each video frame and the video soundtrack.

2. Summary Section

  • Aggregates all frame-level tags into a single overview.

  • Contains tags that were detected throughout the video frames and those assigned to the entire video.

  • Helps users get a quick summary of key tags without inspecting individual frames.

3. Tag Editing and Creation

  • Users can edit tags assigned to each video frame.

  • If a video frame lacks a tag or is missing from the list, users can create a new tag.

  • Tag editing and creation is done via the Tag Editor, which is accessible by clicking the “+ Tag” button.

4. Tag Search and Navigation

  • Users can search for specific tags within the Tag Log.

  • Navigation buttons (Previous and Next) allow users to move between occurrences of the searched tag efficiently.

5. Exporting Tag Data

  • Users can export both the Summary and Frame-Level Tags as an Excel spreadsheet by clicking the Download button.

  • The exported file contains all visible tag information from the Tag Log interface.

  • Additionally, the downloaded file includes probability scores for each detected tag, offering insight into the confidence level of the AI-generated classifications.

Did this answer your question?