Audio & Video Socializing
Holla
How Holla Group secures real-time 1v1 video chat with millisecond stream-cutting — while powering smarter user matching.
98%
Detection Precision
95%
Attribute Tagging Precision
ms
Real-Time Interception
The Customer
A pioneer in
cross-border video social.
Holla Group is a leading operator of real-time video social platforms, with a portfolio of globally popular 1v1 random video chat applications including HOLLA, Monkey, Three, and hay.
As a pioneer in the cross-border video social space, Holla Group is dedicated to providing a safe and healthy real-time interactive experience for users worldwide.
At a Glance
IndustryAudio & Video Socializing
AppsHOLLA · Monkey · Three · hay
FormatReal-time 1v1 video chat
DeepCleer ProductsLive Stream · Image Moderation
The Challenge
Complex risk —
at global scale.
CHALLENGE 01
High-Risk Real-Time Compliance
Real-time video streams are highly susceptible to content containing minors, NSFW, or other explicit material that severely violates laws (such as COPPA) and platform policies — requiring ultra-real-time detection and enforcement.
CHALLENGE 02
User Matching Precision
To enhance in-app user matching, the recommendation algorithm requires accurate image attribute tagging — gender, age, and image quality — as structured input.
CHALLENGE 03
High Concurrency, Low Latency
Video-frame auditing must complete within milliseconds to ensure non-compliant content is intercepted instantly via stream cutting, without disrupting the experience for legitimate users.
The Solution
A closed-loop strategy
tuned to the platform.
01
Full-Volume Video Frame Auditing
All video frames collected client-side are sent for auditing. DeepCleer's image recognition precisely identifies risks across political content, NSFW, terrorism, contraband, and faces.
02
Hierarchical Real-Time Enforcement
Images judged REJECT trigger instant "real-time stream cutting" — blocking the broadcast and banning the account, followed by manual review. Severe-violation REVIEW images route to efficient manual review so no high-risk content leaks.
03
Business Recommendation Tagging
Visual recognition identifies and tags gender, age, and image attributes in normal video scenes. This structured data is returned via API to optimize user-matching algorithms — turning safety into business value.
More Customer Stories
Trusted across AI, social & gaming.
Ready to Secure
Your Platform?
Get a personalized demo with your content types and use cases.
