DocsICP FilingConsole
官方文档
首页

Features

更新时间：
复制 MD 格式
产品详情
我的收藏

This topic describes the features of Content Moderation to help you better understand and use the product.

Content Moderation API

Application ScenarioDescription
Image moderationDetects violations or inappropriate information in images. The following scenarios are supported:
  • Smart pornography detection
  • Detection of terrorist and political content
  • Text and image violation detection
  • QR code detection
  • Adverse scene detection (no content, Picture-in-Picture (PiP), smoking, in-car live streaming)
  • Logo detection
Video moderationDetects violations or inappropriate information in videos. The following scenarios are supported:
  • Smart pornography detection
  • Detection of terrorist and political content
  • Text and image violation detection
  • Adverse scene detection (such as black screens and white screens)
  • Logo detection
  • Voice moderation
Text spam detectionDetects violations or inappropriate content in text. The following scenarios are supported:
  • Ad detection
  • Detection of terrorist and political content
  • Abuse detection
  • Pornography detection
  • Spam detection
  • Meaningless content detection
  • Contraband detection
  • Adverse scene detection (supports detection of scenarios such as money worship, fan culture, negative emotions, and negative inducement)
  • Custom keyword detection
Voice spam detectionDetects violations or inappropriate content in voice data. The following scenarios are supported:
  • Ad detection
  • Detection of terrorist and political content
  • Abuse detection
  • Pornography detection
  • Spam detection
  • Meaningless content detection
  • Contraband detection
  • Adverse scene detection (supports detection of scenarios such as money worship, fan culture, negative emotions, and negative inducement)
  • Custom keyword detection
File spam detectionDetects violations or inappropriate content in files. The following scenarios are supported:
  • porn: Pornography detection
  • ad: Ads
  • terrorism: Terrorist and political content
  • sface: Sensitive faces
  • qrcode: QR codes
  • live: Adverse scenes (no content, PiP, smoking, and in-car live streaming in images; black screens and white screens in videos)
  • logo: Logo detection
  • ocr: OCR for text and images
Webpage spam detectionDetects violations or inappropriate content on webpages. The following scenarios are supported:
  • porn: Pornography detection
  • ad: Ads
  • terrorism: Terrorist and political content
  • live: Adverse scenes (no content, PiP, smoking, and in-car live streaming in images; black screens and white screens in videos)
OCR for text and imagesRecognizes various text (unstructured information) in images. The following capabilities are included:
  • General OCR for text and images
  • General OCR for text and images (high-precision)
Facial recognitionFacial recognition includes the following capabilities:
  • Facial attribute detection
  • Face comparison
  • Sensitive face detection in images
  • Sensitive face detection in videos
Image label recognitionDetects the main entities in an image and returns the corresponding labels.
Similar image searchSearches a user-defined image library for the N images that are most similar to a given image.
Video quality detectionDetects quality issues in a given video file, such as blur, splicing, low brightness, black screens, and frozen frames.
Structured card and certificate OCRStructured card and certificate OCR recognizes text in certificate images and returns the text in a structured format.
Liveness and screen replay detectionDetects whether a live object (mainly a human face) in an image is from a direct capture or a screen replay.
Custom face searchPerforms custom face searches as needed.

OSS violation detection

Application ScenarioDescription
Detects whether images, videos, and voice data in Object Storage Service (OSS) contain violations such as pornography and political contentDetects violations in OSS buckets. Supports automatic detection of incremental content and manual scanning of existing content.