Overview

Content Moderation leverages Deep Learning technology and benefits from Alibaba's years of Big Data analysis to provide accurate monitoring of pictures, video, text, voice and other multimedia content. Not only does Content Moderation help users to reduce adult, violence, contraband, drugs and other illegal or inappropriate content, but can also minimize spam advertising and other user experience pain points.

Benefits

• Fast, Accurate, and Reliable Constant automated moderation responses in less than 0.1 second with an accuracy rate higher than 95 percent.
• Comprehensive Service Readily recognizes adverse images, videos, text, and audio dealing with illicit behaviors such as violence, contraband, drugs, weapons, extremism and profanity.
• Proven Performance Daily access to billions of images, videos, text, and audio with highly scalable, deep learning technology developed by Alibaba.
• Customizable Customize models according to your specific requirements. Continually improves recognition based on new data to expand its capabilities.

Features

Image Moderation

• Image Moderation v2.0 baselineCheck Support the detection of red line content in images, such as pornography, sexiness, violence, prohibition, flags, inappropriate, abusive, and special elements, including the content of the images and the text content in the images (supporting 18 languages such as Chinese, English, French, Russian, Japanese, Arabic, etc.).
• Adult Utilize an API to detect and filter images containing adult content
• Violence & Contraband Utilize an API to detect and filter images containing violence or contraband content
• undesirable Detects undesirable scenes, such as black screen, black edge, dark image, Picture-in-Picture (PiP), smoking, and live streaming while driving, in images.
• Logo Utilize an API to detect logos within images
• OCR OCR is used for extracting text from images
• Face Detect and Search Supports detection of faces in images, comparison of facial features, and retrieval of facial features.

Video Moderation

• baselineCheck_global Detect whether there are illegal or inappropriate content in video files, and support the detection of images and voice in videos. It is recommended to perform this detection on all video files that involve open public network access.
• Video Live Stream Detection Detect whether there is any illegal or inappropriate content in the live video stream, and support the detection of live video images and voice. It is recommended to perform this detection on all video live streams involving open public network access.
• Adult Utilize an API to detect and filter videos containing adult content
• Violence & Contraband Utilize an API to detect and filter videos containing violence or contraband content
• undesirable Detects undesirable scenes in a video.
• Logo Utilize an API to detect logos from videos

Text Moderation

• comment multilingual global Adopting a separate strategy and labeling system can effectively identify text content such as pornography, contraband, advertising, verbal abuse, violation of advertising laws, and involving prohibited items. Provide more functional features, simplify business usage, and assist in manual auditing. Supports 38 language types including Chinese, English, French, Thai, Japanese, Korean, Russian, Portuguese, and Arabic.
• Text Moderation Utilize an API to detect and filter text containing adult, abusive, spam content. Current service can only support Chinese and English language.

Voice Moderation

• Voice Moderation 2.0 audio_multilingual_global Voice Moderation V2.0 service upgrades the content security core engine to provide audit services for business scenarios such as image and text sharing, game streaming, and live courses. It identifies content or elements that violate network content dissemination regulations, affect platform order, and user experience, and provides rich content risk labels.