Aestron is an innovative, open, full-stack, all-scenario audio and video solutions provider, focusing on real-time interaction technologies. Leveraging artificial intelligence, big data, and real-time communication technologies, Aestron helps enterprises quickly build real-time audio and video calls, interactive live streaming, and real-time messaging applications. Aestron currently helps serve more than 400 million users worldwide, covering more than 150 countries and regions.
Accurately identifies texts in various scenarios involving political figures, political events, religion, reactionism, separatism, terrorism, etc.
Accurately identifies pornographic contents such as obscenity, smut, seduction, and sexting; supports hardcore pornography and softcore pornography classification
Accurately identify abusive behaviors, including insulting, abusing, denigrating, etc.
Accurately identifies illegal contents such as gambling, knives and guns, drugs, counterfeiting, counterfeiting, etc.
Accurately identifies illegal spam advertising contents which may contain WeChat accounts, phone numbers, QQ accounts, QR codes, etc.
Accurately identifies spam contents such as spam posting, flooding, and irrelevant content posting.
Public messages in live streaming, overlaid comments on videos, IM messages, personal web pages, self-media articles, product reviews, and forum comments.
Based on the sensitive word datasets from various regions of the world and the NLP technology iteration over the years, Aestron has built a comprehensive modeling system to avoid false positives and false negatives of a single model of false, allowing it to achieve high recognition accuracy.
Aestron offers real-time public opinion monitoring, dynamic tracking of violations, second-level update of sensitive word datasets, and rapid product iterations and upgrades.
Aestron supports hundreds of billions of samples, hundreds of thousands of sensitive words, and tens of thousands of sensitive word variants, dozens of semantic labels and provides efficient recognition of various violations involving politics, pornography, prohibited content, spam, and other illegal content.
Aestron deploys the text moderation service in multiple clusters around the world with auto-scaling and processes tens of billions of text recognition requests every day.
Accurately identifies political figures, military uniforms, spoof cartoons, reactionary figures, and other politics-related contents in images
Accurately identifies pornography, pornographic cartoons, child nudity, and other pornographic contents in images; support the quick classification of hardcore pornographic images, softcore pornographic image, sexy images, and normal images.
Accurately identifies various violent scenes involving terrorist organizations, cults, guns and knives, excessive bleeding, and rioting.
Accurately identifies prohibited behaviors such as smoking, taking drugs, gambling, and kissing.
Accurately identifies illegal spam advertising contents which may contain WeChat accounts, phone numbers, QQ accounts, Weibo accounts, URLs, QR codes, and other variants.
Accurately identifies minors in various scenes and estimates the age distribution of the minors.
Personal web page, avatars and background images, cover images for livestreams and videos, and images in forum comments, product comments, and social messages.
Based on deep fusion of classification models, detection models, and OCR models, Aestron has built a comprehensive modeling system to avoid false positives and false negatives of a single model, allowing it to achieve high recognition accuracy.
Aestron offers real-time public opinion monitoring, dynamic tracking of violations, day-level update of model updates, and rapid product iterations and upgrades.
Aestron provides comprehensive recognition of violence, terrorism, pornography, spam, advertisements and theirs variants in images, significantly reducing the risk of violating the rules by the governments.
Aestron deploys the image moderation service in multiple clusters around the world with auto-scaling and processes billions of image recognition requests every day.
Accurately identifies political figures, political events, reactionism, separatism, terrorism, and other politics-related contents in audio files.
Accurately identifies audio that involves moaning, ASMR, rapping, and other illegal contents based on a in-house developed moaning sound recognition technology.
Accurately identifies pornographic contents such as obscenity, smut, seduction, phone sex, etc.
Accurately identifies abusive behaviors, including insulting, abusing, denigrating, etc.
Effectively identifies illegal content involving pornography, politics, abuse, and advertising in multiple languages in audio files.
Online education, chat room, live voice/video channel, audio introduction for personal web pages, audio files in IM, FM radio, custom voices, audio and video recordings
Based on deep fusion of classification models, detection models, and OCR models, Aestron builds a comprehensive modeling system to avoid false positive and false negative recognition of a single model of false, and achieve high recognition accuracy.
Aestron offers rich functions such as speech recognition, semantic labeling, moaning recognition, noise recognition, multilingual audio recognition to meet the needs of illegal audio detection and personalized audio recommendation in various scenarios such as live streaming, custom voices, FM radio, etc.
Aestron deploys the audio moderation service in multiple clusters around the world with auto-scaling, and processes millions of hours of audio every day.
Accurately identifies political figures, military uniforms, reactionary figures, spoof videos, and other politics-related contents in videos
Accurately identifies pornography, sex, vulgarity, dirty, child nudity and other pornographic contents in videos.
Accurately identifies various violent scenes involving excessive blood, symbols of terrorism, guns and knives, and rioting in videos.
Accurately identifies prohibited behaviors such as smoking, taking drugs, gambling, and kissing in videos.
Accurately identifies illegal spam advertising contents which may contain WeChat accounts, phone numbers, QQ accounts, Weibo accounts, URLs, QR codes, and other variants.
Accurately identifies watermark on videos such as brand logos and government-related logos.
Live streaming, short-form videos, video hosting websites, e-commerce product introductions, and group video chats.
Based on deep fusion of classification models, detection models, and OCR models, Aestron builds a comprehensive modeling system to avoid false positive and false negative recognition of a single model of false, and achieve high recognition accuracy.
Aestron offers real-time public opinion monitoring, dynamic tracking of violations, day-level update of model updates, and rapid product iterations and upgrades.
Aestron provides comprehensive moderation of images, audios, subtitles, texts, and recognition of illegal contents such as politics, violence, terrorism, pornography, spam, and watermarks.
Aestron deploys the video moderation service in multiple clusters around the world with auto-scaling and processes tens of millions of video recognition requests every day.