Image Moderation Solution
Our image moderation solution can detect the following classes:
Porn
Descriptions of sexual intercourse that include physiological details, descriptions of genitals, and actions with them in any context, artistic or medical.
Erotica
Descriptions of sex or someone's appearance that are informational in nature and don't mention gross physiological details.
Sex toys
The image contains sex toys (like vibrators, handcuffs, and BDSM gear), sex shop ads, lubricants, condoms, dietary supplements, or vitamins and creams for improving sexual experiences
Alcohol
The picture shows alcoholic beverages or their consumption.
Tobacco
The picture shows tobacco products, electronic cigarettes, cigars, hookahs, or smoking
Drugs
Ads or propaganda of narcotic substances.
Casino
The image shows gambling tables, slot machines, online casino and gambling apps on a phone/tablet/computer, and slot machine screensavers.
Violence
Scenes of violence, brutality, fights, blood, wounds, dismemberment, or horrified faces
Shock-content
Visualization of diseases and anatomical details in an unpleasant or frightening context, cosmetological or medical injections (involving syringes), and dental manipulations.
Offensive content
Insults addressed at any people or material objects.
Clickbait
A message that excites and arouses curiosity but doesn't give complete information.
Our solutions are applied in the following industries
We support various languages
But we flexibly adapt all our
solutions to your needs and take into account the features of your service
Our platform in numbers
- 35ClassesFlexibilityCustomizable moderation classesfor brand-appropriate responses
- 30ModelsConstant fine-tuningOur crowd force can handle anygranularity of detail, and our models continue to learn from incoming user data
- 95%AccuracyHigh accuracy in nonstandard casesWe apply a multitude of QA methods to both our human-in-the-loop moderators and our ML models to bring you the most accurate and high-quality results
- >15MCases per dayScalabilityWe can process any volume of data, including spikes caused by spam attacks, breaking news, flash mobs, or organic growth
- 15 minAverage speed of crowd checkLarge and fast crowdWe have over 2M workers distributed over the globe, so manual check are available 24/7
- 200msFrom request to resultReal-time moderationFor rapidly growing, large-scale, and fast-paced businesses
Other solutions
Text
Detect unpleasant or rude communication or mentions of prohibited products and services.
Show more →Video
Check videos for prohibited scenes and content that might interfere with a pleasant viewing experience.
Show more →Audio
Identify unwanted content in audio tracks, check the sound quality and signal.
Show more →