Configure a full scan task

Background

For more information about the limitations of OSS compliance detection, such as supported OSS buckets, detection scenes, and file formats, see Limitations.

Prerequisites

• You have activated OSS and granted Content Moderation permission to read your OSS buckets. For more information, see Authorize Content Moderation to access OSS buckets.

• You have configured callback notifications. For more information, see Configure message notifications.

Step 1: Create a full scan task

1. Log on to the Content Moderation console.

2. In the left-side navigation pane, choose OSS Scan > Existing Data Scan.

3. Click Create scan task to configure a full scan task.

4. In the setup wizard, configure the following settings.

   1. Select Bucket.

      From the Unselected box, select the buckets that you want to scan and add them to the Selected box. Then, click Next.

   2. Set filter conditions.

      Configuration instructions:

      • Filter or exclude: By default, all files in the selected buckets are scanned. If you want to scan only specific files or exclude specific files from the scan, click Set filter conditions to the right of a bucket. In the Filter Conditions dialog box, configure the filter conditions and click OK.

        Table 1. Filter condition parameters

        Parameter	Description
        Filter condition	You can either Include specified paths in the scan or Exclude specified paths from the scan. Select only one option.
        Path	Enter the path prefix to include or exclude. Click Add to add multiple paths. You can add up to 10 paths. Example (using the Include condition): Add img/test_ to scan only files in the bucket whose keys start with the img/test_ prefix. If the files you want to scan are in a specific directory, you can prepend the directory path to the file prefix. For example, if the files are in the img/202011 directory and have the test_ prefix, you can add img/202011/test_ as the path.

      • Time range (required): Set a File upload time range. Content Moderation scans only files uploaded within the specified time range.

      Then, click Next.

   3. Configure detection scenes.

      OSS compliance detection provides eight recommended configurations. Each configuration includes main scene categories (Image, Video, Audio) and specific sub-scenarios, which vary for each configuration. If the recommended configurations do not meet your business needs, you can enable or disable the main categories but cannot modify the sub-scenarios.

      Then, click Next.

   4. Configure scan scope.

      The scan scope settings depend on your selected detection scenes. For example, if you select only the Image category, only the image configuration settings are displayed. If the current configuration does not meet your business needs, you can customize the scan scope.

      Note

      You must select at least one detection scene. You must enable scanning for at least one of the Image, Video, or Audio categories.

      Table 2. Scan configuration parameters

      Scan scope	Parameter	Description
      Image	Image scan limit	The default limit is 10,000 images per bucket. If you set a scan limit, the scan stops after the limit is reached. This can expose non-compliant images. We recommend not setting a scan limit.
      	Detect files without suffixes	If you enable Detect Files Without Suffixes, image files that do not have a file suffix are scanned. This reduces detection speed. The system determines whether a file is an image based on the content-type in the HTTP header. Supported content-type values include image/jpg, image/jpeg, image/png, image/gif, image/bmp, and image/webp.
      Video	Video scan limit	The default limit is 1,000 videos per bucket. If you set a scan limit, the scan stops after the limit is reached. This can expose non-compliant videos. We recommend not setting a scan limit.
      	Frame capture frequency	The default is 1 frame per second. You can set the interval between captured frames from 1 to 60 seconds. A smaller interval (higher frequency) results in higher detection accuracy.
      	Maximum frames per video	The default limit is 200 frames per video. You can set the maximum number of frames captured per video. The value can be from 5 to 20,000.
      	Maximum size per video	The default limit is 500 MB per video. You can set the maximum size for a single video. The value can range from 1 to 2,048 MB. Any part of the video that exceeds this size is not scanned.
      Audio	Audio scan limit	The default limit is 1,000 audio files per bucket. If you set a scan limit, the scan stops after the limit is reached. This can expose non-compliant audio content. We recommend not setting a scan limit.
      	Maximum audio file size	The default limit is 200 MB per audio file. You can set the maximum size for a single audio file. The value can range from 1 to 2,048 MB. Any part of the audio file that exceeds this size is not scanned.

      Then, click Next.

   5. Optional: Other. In the Callback notifications section, you can select an existing Callback notification scheme to receive the scan results.

      If you have not yet created a callback notification plan, click Add Notification Plan to create one. For more information, see Configure message notifications.

5. At the bottom of the page, click Submit.

   OSS compliance detection estimates the maximum scan fee based on your configuration. You can choose the pay-as-you-go billing method or use a resource plan to cover the cost, based on your needs.

   Then, click OK.

   After you submit the full scan task, the scan starts immediately. The scan duration depends on the number of existing files and the enabled detection scenes. You can view the status of the task in the full scan task list.

   OSS compliance detection provides the following features. You can perform the following operations:

   • Pause scan

     To pause a running task, click Pause Scan in the Actions column. You can resume a paused task within 7 days.

   • Data statistics

     To view the API call statistics for the current task, click Data Statistics in the Actions column. On the OSS Violation Detection API Calls page, you can view the call volume for the last 7 days. You can also customize the query by setting conditions such as the scan time and task.

   • Stop scan

     To stop a running task, click Stop Scan in the Actions column. After you stop a task, the remaining files will not be scanned.

   • Scan results

     To view the scan results, choose More > Scan Results in the Actions column if the task is running, or click Scan Results if the task is complete. You can filter the results by time range or task.

   • View configuration

     To view the configuration of the current task, choose More > View Configuration in the Actions column if the task is running, or click View Configuration if the task is complete. The details are displayed in the Task Configuration Details panel.

   • Usage notes

     To view the resource usage for the current task, choose More > Usage Notes in the Actions column. The details are displayed in the Task Usage Notes panel.

Step 2: View scan results

After a full scan task completes, you can view the results and perform manual reviews in the Content Moderation console.

1. On the Full Scan page, view the Task overview, Task Status, and Task results for your task.

2. In the Actions column, click Scan Results to view the scan results and perform a manual review.

   By default, the scan results and processed non-compliant content from the last 7 days are displayed. You can filter the results by scan scope and export them. When you export results, all records are included if there are fewer than 50; otherwise, only non-compliant and suspected non-compliant records are exported.

   Note

   You can click an image or video in the scan results to view its Details, including the file creation time, key (the object key), and the bucket where it is stored.

   If a scan result is incorrect, you can manually correct it. You can perform the following operations:

   • Mark as Violation and Delete

     Click Mark as Violation and Delete to delete one or more selected items from both the Content Moderation console and the OSS bucket.

   • Mark as Normal and Ignore

     Click Mark as Normal and Ignore to dismiss the detection result. Ignored items are removed from the console but remain in the OSS bucket. You can select one or more items to ignore.

   • Normal and unfreeze

     If the automatic freezing feature is enabled, you can also select a frozen image or video and click Normal and unfreeze to unfreeze it.

Step 3: View statistics

After a full scan task completes, you can view its data statistics in the Content Moderation console.

1. On the Full Scan page, click Data Statistics in the Actions column.

2. On the Data Statistics page, click the Image, Video, and Audio tabs to view scan statistics for the last 7 days.

   The following table describes the supported data statistics.

   Query object	Supported statistics
   Image	Total images: Total scanned images. Pornography detection volume: The total number of results (including non-compliant, suspected, and normal) from the pornography detection scene. Terrorism and politics detection volume: The total number of results (including non-compliant, suspected, and normal) from the terrorism and politics detection scene. Ad detection volume: The total number of results (including non-compliant, suspected, and normal) from the ad detection scene. Adverse scene detection volume: The total number of results (including non-compliant, suspected, and normal) from the adverse scene detection.
   Video	Total videos: Total scanned videos. Total frames: Total captured video frames. Number of videos for pornography detection: The total number of video results (including non-compliant, suspected, and normal) from the pornography detection scene. Number of videos for terrorism and politics detection: The total number of video results (including non-compliant, suspected, and normal) from the terrorism and politics detection scene. Ad detection volume: The total number of video results (including non-compliant, suspected, and normal) from the ad detection scene. Adverse scene detection volume: The total number of video results (including non-compliant, suspected, and normal) from the adverse scene detection. Audio spam detection volume: The total number of audio results (including non-compliant, suspected, and normal) from audio tracks in scanned videos.
   Audio	Total audio files: The total number of scanned audio files. Total audio duration: Total duration of scanned audio files. Audio spam detection volume: The total number of results (including non-compliant, suspected, and normal) from audio spam detection.