Aestron AI Content Moderation-HUAWEI CLOUD

3 Highlights 3 Highlights

Visual Recognition: Intelligent Image Moderation System

AI intelligent algorithms can accurately and efficiently identify multiple types of image or video content, such as pornography, violent and terrorism, blood, politics, juveniles, etc. The system uses a variety of models to comprehensively identify, effectively control the cost of manual review, and improve the efficiency of industry content review.
Audio Recognition: Intelligent Audio Moderation System

Covers ASR, language classification, voiceprint recognition and other fields, supports the recognition of Chinese, English, Indonesian, Hindi, Arabic, Japanese and other languages, and accurately identifies risks such as pornography, political and religious sensitivity, moan and insult content.
Text Recognition: Intelligent Text Moderation System

Based on the abundant sensitive vocabulary and policy rule model, the system efficiently detects and recognizes illegal content, covering Chinese, English, Russian, Arabic, Hindi, Indonesian, Malay, Thai, Vietnamese, Spanish, German, and Portuguese, etc.

Aestron AI Content Moderation

Company introduction

Aestron is a leading global internet technical service provider, focusing on artificial intelligence and cloud communication technology. We empower our enterprise customers with up-to-date technologies including real-time AI content moderation, global messaging and content delivery network, and help reduce costs while improve efficiency during their business globalization.

Industry challenge

Social industry

Personal information like nicknames, avatars and dynamics, private chat, group chat, posting, comment and other links, containing content risks such as pornography, insults, violent and terrorism, political-related, advertising diversion, etc.

Pan-entertainment industry

Live, videos, broadcasts, live screens, bullet screens, comments, covers and other links, containing content risks such as politics, pornography, insults, violent and terrorism, ad diversion, etc.

Game industry

Game players’ nicknames, avatars, public screen chats, containing content risks such as politics, pornography, violence and terrorism, ad diversion, etc.

E-commerce industry

Personal information, reviews, product descriptions, good things sharing, containing content risks such as politics, pornography, insults, violent terrorism, ad diversion, etc.

Core Features

High-precision and High-recall

Basing on massive data and leading vision technology, the recognition accuracy and recall of various models are industry-leading.

Multi-modal Fusion and Multi-tag Combination

Image and audio fusion to identify video tags, provide risk recognition services with high accuracy, and cover vertical domains such as pornography, terrorism, politically sensitive figures, etc.

Fast Iterative Update of Model

Ten-thousand-level sensitive library data is updated daily, continuously iterating the effect of image recognizing capabilities, and can identify new types of complex and illegal images

Stable Service and Real-time Monitoring.

According to the request volume under different business scenarios, the capacity can be flexibly expanded to ensure the high QPS of the service, and the real-time monitoring system is configured to monitor alarms and log viewing at any time.

Some advantages to the target audience

Monitoring & Response

Performs real-time feature retrieval and risk interception according to the latest instructions of the Office of the Cyberspace Administration and the monitoring of public opinion content.

Strong Technical Support

Leverages expertise of more than 1200 technicians, about 100 of whom are AI experts and engineers.

Solution Customization

Provides in-depth consulting services and cost-effective solutions, and supports custom sensitive dataset and feature recognition models, etc.