Search library management

更新时间:
复制 MD 格式

A search library helps you search your media assets, such as videos, audio, and images, more efficiently and accurately. When you upload a media asset to a search library, the system automatically analyzes its content and saves the results. You can then use the corresponding search features or APIs to quickly find the assets you need.

Billing

  • You are charged based on the duration of the videos analyzed and the volume of video storage used in the search library. To avoid unnecessary costs, disable the service when not in use. For more information about billing, see intelligent search.

  • Billing starts automatically as soon as you enable an index. If you no longer need the search functionality, make sure to disable the index to stop charges.

  • If you manually disable an index, the corresponding data in your intelligent search library is cleared, and you can no longer perform that type of search.

  • After your enterprise subscription (including trial and paid versions) expires, your search library enters a 7-day protection period. After the protection period ends, your intelligent search library is automatically cleared, and all search functions become unavailable.

Notes

  • An active index ensures that the system automatically analyzes new videos and images uploaded to your media asset library.

  • To ensure that new media assets are correctly analyzed and indexed, we recommend keeping the index status switch turned on.

Index types

A search library supports three index types: smart tag, face search, and large model search.

  • Smart tag: Enables tag-based search for videos and images. You can search for analyzed media assets by using AI-generated category tags, text recognized in videos, and transcribed speech.

  • Face search: Enables face search for videos and images. You can search for analyzed media assets by using a person's name or a face image.

  • Large model search: Allows you to search for videos and images by using natural language descriptions. This feature uses a large model to map visual and text content to a unified high-dimensional space, preventing semantic loss or inconsistencies that can occur when converting video to text. It also converts your search text into a high-dimensional vector without word segmentation. For example, you can search for "a singer riding a bicycle" as a single concept, making your searches more efficient and intuitive.

Create a search library

Using the console

  1. Log on to the Intelligent Media Service console.

  2. In the upper-left corner, select the region where your service is located.

  3. In the navigation pane on the left, choose Media Asset Management > Search Management > Search Library.

  4. Click Create Search Library.

    A search library supports three index types: smart tag, face search, and large model search. You can enable the indexes you need.

    Index type

    Description

    Smart tag search

    Click to enable smart tag search.

    A confirmation dialog box with the title I am aware of the billing rules for search appears, stating that billing starts after you enable the feature. If you disable it, the smart tag index in the search library is automatically cleared. After your enterprise subscription expires, you have a 7-day protection period for your search library, after which it will be automatically cleared. Select the I am aware of the billing rules for search checkbox and click OK.

    Face search

    Click to enable face search. You can select either a shared instance or a dedicated instance based on your needs.

    • Shared instance: Uses computing resources that are shared with other users. This option is suitable for testing and trial purposes.

    • Dedicated instance: Provides exclusive computing resources that are not shared with other users. This option is suitable for commercial use.

      In the Search Specifications dialog box, select a shared instance or dedicated instance specification. Select the I am aware of the billing rules for search checkbox, and click OK. Instance creation takes about 2 to 3 minutes. Billing starts after face search is enabled.

    Large model search

    Click to enable large model search. You can select either a shared instance or a dedicated instance.

    In the Search Specifications dialog box, a shared instance supports up to 1,000 hours and uses computing resources shared with other users, making it suitable for testing and trial purposes. A dedicated instance provides exclusive computing resources and is suitable for commercial use. You can select a duration from the drop-down list (for example, up to 10,000 hours). Billing starts after large model search is enabled. Instance creation takes about 2 to 3 minutes. After selecting a specification, click OK.

  5. Click Submit to complete the creation.

Using OpenAPI

  1. Call the Create a search library operation to create a search library with a specified name.

  2. Call the Create a search index operation to create an index in the search library.

  3. Set the IndexType parameter based on the type of index you want to create:

    1. Smart tag search: aiLabel

    2. Large model search: mm

    3. Face search: face

Activate index

If an index is inactive, you can activate it by selecting the specific search library and clicking Activate Index. Activation means the system will automatically analyze new content added to the media asset library and save the analysis results to the search library. You can then find media assets by using features such as smart tag search, face search, and large model search.

A Billing Notice dialog box appears, informing you that charges will apply after activation. Click OK to activate the index.

Upload media assets

Using the console

When you upload media assets in the console, you must select a specific search library. For detailed instructions on uploading, see Console Upload.

On the right side of the upload page, click the Batch Modify Search Library drop-down list and select the Default Search Library or a custom search library you have created (for example, test). You can also click Add Search Library to create a new one.

After the upload is complete, the media asset is automatically analyzed. You can search for the asset after the analysis is finished. To check the analysis status, click Manage in the media asset list to open the details page and view the asset's index status.

Using OpenAPI

Call the Register a media asset operation to upload a media asset. In the Register a media asset request, set the SearchLibName parameter. For example: {"RegisterConfig":{"SearchLibName":"test"}}. If you do not specify this parameter, the system automatically uses the default search library.

View analysis results

After you create a search library and activate its indexes, the system can intelligently analyze any media assets uploaded to it. In the Actions column, click Media Asset Management to check if the index status is active. Once active, the system analyzes and indexes the media assets. The analysis results are automatically saved to the search library, allowing you to find resources using search features that correspond to the index type.

On the media asset management page, click the filter icon in the Index status column header. You can filter assets by index type (smart tag, face search, or large model search) and by their processing status (Success, Failed, or Running).

Search media assets

Using the console

If a media asset has been analyzed and indexed, you can go to the intelligent search page to find it by using smart tag search, face search, or large model search.

Using OpenAPI

  • Use the smart tag text search API to search for media assets. This is a two-stage process:

    1. First, perform a coarse search.

    2. Then, perform a fine-grained search. The response returns a list of specific clips that match within a particular media asset.

  • Use the large model search API to search for media assets. The response returns a list of matching media assets and the matching clips within each asset.

  • Use the face image search API to find clips containing a specific face within a media asset. This process has two stages:

    1. Call the search for media assets by face image API (coarse search). Provide a face image to search for media assets containing that person.

    2. Call the search for media clips by face image API (fine-grained search). Based on the coarse search results, this operation returns specific clips where the face appears.

Manage search libraries

You can manage your created search libraries, including modifying their configurations or deleting them.

  • Modify index configuration

    In the Actions column, click Configure to modify the index configuration.

  • Delete a search library

    1. Before you can delete a search library, you must first delete the index.

      Important

      Deleting an index permanently removes all data stored in the search library and releases the instance resources. This action cannot be undone.

      On the search library management page, select the target search library and click the Delete Index button in the batch operations area at the bottom.

    2. Click the Delete button to delete the search library. This action permanently removes all media assets and index content within the library.

      This action cannot be undone. In the confirmation dialog box, enter I am aware and confirm the deletion to confirm, and then click OK.

FAQ

What should I do if the index status is "Failed"?

Related APIs