Image Moderation Enhanced Edition console

When you first use the service, log on to the console and go to the Rule Configuration page to review the default detection scope settings.

Scenarios

If your business has the following needs, use the Image Moderation Enhanced Edition console to customize detection rules, query results, and view usage statistics.

Prerequisites

Adjust risk detection scope

Adjust the detection scope and risk items to meet your business requirements.

1. Log on to the Content Moderation console.

2. In the left navigation pane, choose Machine Moderation Enhanced Edition > Image Moderation > Rules.

3. On the Rules Management tab, find the service that you want to manage, for example, Baseline Check (baselineCheck). Click Settings in the Actions column.

4. On the Detection Scope page, select a detection category to adjust. This example uses Prohibited Content Detection.

   1. On the Prohibited Content Detection tab, review the current default settings in the Detection Scope Configuration section. By default, four checks are enabled, and the service returns a corresponding label if a match is found.

      The default configuration includes the following four checks: contraband_drug (images suspected of containing drugs or medications), contraband_drug_tii (text in images suspected of describing prohibited drugs), contraband_gamble (images suspected of containing gambling-related items), and contraband_gamble_tii (text in images suspected of describing gambling activities). The default medium risk score range is 50.00–79.99, and the high risk score range is 80.00–100.00. The detection status for all checks is On.

   2. Click Edit to enter edit mode and modify the On or Off. For example, turn off the detection switch for the third item.

      The third item is contraband_gamble (images suspected of containing gambling-related items). After turning it off, click Save.

   3. You can also adjust the Medium risk score and High risk score thresholds to define the returned risk level.

      Note

      The risk level is determined based on the following rules:

      1. If a detected risk label's confidence score is in the high-risk range, the result is "high risk".

      2. If a detected risk label's confidence score is in the medium risk score range, the result is "medium risk".

      3. If a detected risk label's confidence score is below the start of the medium risk score range, the result is "low risk".

      4. If multiple labels with different risk levels are detected, the highest risk level is returned.

      5. If no risk labels are detected, the result is "no risk".

      6. If an image matches an entry in a custom blocklist, the result is "high risk".

   4. Click Save. The new configuration takes effect in about 2 to 5 minutes.

Configure differentiated scopes for business scenarios

Copy a service and configure different risk detection scopes for different business scenarios.

1. Log on to the Content Moderation console.

2. In the left navigation pane, choose Machine Moderation Enhanced Edition > Image Moderation > Rules.

3. On the Rules Management tab, copy the service for which you want to set a different detection scope. This example uses Baseline Check (baselineCheck).

   1. In the service list, find the Baseline Check service and click Copy in the Actions column.

   2. In the Copy Service panel, enter a Service name and Service description.

      The Service field displays the current service name. You can customize the Service Name (up to 32 characters) and Service Description (up to 64 characters) fields.

   3. Click Create to save the copied service information. The new service is available for API calls in 1 to 2 minutes.

   4. After the service is created, you can set and edit rules for the copied baselineCheck_01 service. By calling the baselineCheck_01 and original baselineCheck services separately, you can meet the needs of different business scenarios that require different risk detection scopes.

Targeted and emergency moderation

Configure a custom image library to flag specific risky images. If a user-uploaded image matches an entry in the library, the service returns a risk label.

1. Log on to the Content Moderation console.

2. Manage your image libraries before configuring them for custom detection. If an existing library already meets your business requirements, you can skip this step.

   Note

   Each account can create up to 10 image libraries, with a total limit of 100,000 images across all libraries.

   • Create an image library and upload images

     1. In the left navigation pane, choose Machine Moderation Enhanced Edition > Image Moderation > Image Libraries.

     2. Click Create Image Library. On the Create Image Library tab, enter a library name and notes, and then click OK.

     3. Find the library you created and click Image Detail in the Actions column. The image library list includes columns for Image Library Name, Notes, Image Count, and Update Time.

     4. Click Add image. In the Add image dialog box, click Select Image and follow the instructions to upload your images.

        You can upload up to 10 images at a time. Each image must be 4 MB or smaller, and we recommend a resolution of at least 256x256 pixels. The upload list displays the status of up to 10 images. To upload more images, click Clear the list and continue uploading images.

        Supported image formats include PNG, JPG, JPEG, BMP, and WEBP. You can also drag images directly into the upload area.

     5. On the library's details page, view the list of uploaded images. You can also query for and delete images.

        • Query images: Search for images by Image ID or Added At.

        • Delete images: Remove one or more images from the library. Batch deletion is supported.

   • Maintain an existing image library

     1. In the left navigation pane, choose Machine Moderation Enhanced Edition > Image Moderation > Image Libraries.

     2. Find the library you want to maintain. Click Edit in the Actions column to modify its name and notes. Click Image Detail in the Actions column to upload or delete images. The image library list includes columns for Image Library Name, Notes, Image Count, and Update Time.

3. In the left navigation pane, choose Machine Moderation Enhanced Edition > Image Moderation > Rules.

4. On the Rules Management tab, find the service that you want to manage, for example, Baseline Check (baselineCheck). Click Settings in the Actions column.

5. On the Detection Scope page, select a detection category to adjust. This example uses Prohibited Content Detection.

   1. On the Prohibited Content Detection tab, view the current custom library configuration in the Set Labels by Customized Libraries section.

      The page is divided into the Sub-scene Configuration and Custom Library/Vocabulary Configuration sections. The Custom Library/Vocabulary Configuration section displays configuration items for image libraries and vocabularies related to prohibited content detection. For labels ending in _lib (such as contraband_drug_lib), an Image Library drop-down list is available. For labels ending in _tii_lib (such as contraband_drug_tii_lib), a Vocabulary drop-down list is available. You can select a custom image library or vocabulary to trigger the corresponding label upon a match.

   2. Click Edit to enter edit mode, and then select the custom image library that you want to configure.

      For this example, with contraband_drug_lib, select the desired custom image library from the Image Library drop-down list.

   3. Click Save to apply the new custom image library configuration.

      The new configuration takes effect in your production environment in approximately 2 to 5 minutes. If a user-uploaded image matches an image in the configured library, the service returns the "contraband_drug_lib" label.

Exempt trusted images from risk detection

Exempt trusted image libraries from risk detection to prevent false positives on safe content.

1. Log on to the Content Moderation console.

2. In the left navigation pane, choose Machine Moderation Enhanced Edition > Image Moderation > Rules.

3. On the Rules Management tab, click Settings in the Actions column for the relevant service.

4. Select the Exemption Configuration tab to modify the exemption settings.

   1. On the Exemption Configuration tab, view the list of custom image libraries and their exemption status.

      By default, the exemption switches for all libraries are turned off. The Exemption Configuration tab lists the libraries, and you can set the exemption status for each one by using its Exemption Switch.

   2. Click Edit and turn on the switch for the library that you want to exempt.

      After you turn on the switch, the Exemption Switch for the corresponding library shows On. Click Save to apply the configuration.

   3. Click Save to apply the new exemption configuration.

      The exemption library takes effect in approximately 2 to 5 minutes. The Alibaba Cloud Content Moderation service compares input images against the images in the selected libraries. If an algorithm detects a match, the system returns the "nonLabel_lib" label and suppresses other risk labels.

Customize detection for text within images

Enable large models for small services

If you are using a small model service for image moderation, you can enable the large model capabilities with a single click to enhance moderation performance.

1. Log on to the Content Moderation console.

2. In the left navigation pane, choose Machine Moderation Enhanced Edition > Image Moderation > Rules.

3. On the Rules Management tab, find the service that you want to manage, for example, Baseline Check (baselineCheck). Click Settings in the Actions column.

4. On the Detection Scope page, select the Enable LLM Moderation checkbox.

5. In the confirmation dialog box that appears, click OK to enable the large model capabilities.

   Note

   The feature takes effect in 3 to 5 minutes. After you enable this feature, the system returns large model results alongside your existing small model results without affecting them. For more information about the returned results, see VlContent.

Test image moderation online

View usage statistics

Track recent image detection volumes to inform your moderation and governance strategies.

1. Log on to the Content Moderation console.

2. In the left navigation pane, choose Machine Moderation Enhanced Edition > Image Moderation > Dashboards.

3. On the Dashboards page, select a time range to query or export usage data.

   • Query usage: You can view daily or monthly usage statistics. Data is stored for one year, and you can query up to two months of data at a time. The page displays a bar chart of call volume for each service within the selected time range.

   • Export usage: Click the icon in the upper-right corner to export usage data by day or by month.

     The exported report is in Excel format and includes only services with call volumes. The following table describes the fields.

     Field	Description	Unit
     Account UID	UID of the account that exported the data.	N/A
     service	The called detection service.	N/A
     Usage	Total number of calls.	Count
     Date	Date when the statistics were collected.	Day/Month

   • View service hit details: Below the usage statistics, the console details the label hits for each called service, displayed as a daily call volume chart and a label proportion treemap.

     • Call volume column chart: Shows the number of daily requests that hit risk labels versus those that did not.

     • Label proportion treemap: Shows the overall distribution of hit labels, arranged in descending order of proportion. Labels with the same prefix have the same background color.