AI-Powered Virtual Background and Blur
Ensure your representatives look professional wherever they are. Blur the background with one click or replace it with your brand visuals — the person stays clear, the environment is under your control.
Try a mode — the person stays sharp, only the background changes.
Home office, crowded open space, or messy desk — the background can overshadow the professionalism of the call. Bitenta solves this with AI-powered person segmentation: Google MediaPipe's selfie segmentation model separates your representative from the background; you can blur the background or replace it with a chosen image, keeping the person clear. Most importantly: all this happens directly in the representative's browser, on the device's GPU — the camera image is never sent to any server for segmentation. This means both a professional look and complete privacy.
A professional look in three steps
Model Loads in Browser
The MediaPipe segmentation model is loaded into the browser on first use.
Person is Segmented
AI separates the person from the background in real-time (GPU accelerated).
Choose Mode
Blur or virtual background image — one click.
AI runs in your browser
Fast, private and serverless — segmentation happens on your device.
MediaPipe selfie segmentation
Google MediaPipe’s selfie segmentation model separates the person from the background in every frame. With this mask the background is blurred or replaced with an image; the person stays sharp.
Processed in the browser — privacy
All segmentation is done directly in the agent’s browser, on the device’s GPU. The camera image is never sent to any server for segmentation.
GPU acceleration & performance
The model runs with a GPU delegate; canvas buffers are reused frame by frame and dynamic frame-rate adaptation is applied. Low CPU, smooth image.
Blur or brand image
Blur the background or replace it with a preset or organization-specific image. You can switch between modes instantly during a call.
What you can do with a virtual background
MediaPipe Segmentation
Real-time person separation with selfie segmentation model.
Complete Privacy
Image is processed in the browser; does not go to the server for segmentation.
GPU Accelerated
GPU delegate + buffer reuse; low CPU, smooth frame rate.
Background Blur
One click with adjustable blur level.
Virtual Background Image
Ready-made backgrounds or organization-specific images.
Brand Image
Upload your company logo/environment as background.
Instant Switching
Seamless switching between modes during the call.
Works in Browser
Runs on modern desktop and mobile browsers with GPU/WebGL support.
Frequently asked questions
How does background replacement work?
AI-powered person segmentation is used. Google MediaPipe's selfie segmentation model separates the person from the background in the image; you can blur the background or replace it with a chosen image — the person stays clear.
Is my image sent to a server?
No. All processing happens directly in the representative's browser, on the device's GPU. Your camera image is never sent to any server for segmentation — the most secure method for privacy.
Blur or virtual background image?
Both. You can blur the background or replace it with ready-made/organization-specific background images (e.g., office, brand image). Switching between modes during the call is possible.
Can I upload my own background or logo?
Yes. In addition to the ready-made backgrounds in the system, you can upload background images specific to your organization; these are offered to your representatives as options (depending on the plan).
Will it slow down my computer?
Segmentation works with GPU acceleration; frame buffers are reused frame by frame and dynamic frame-rate adaptation is made for performance. It runs smoothly with low CPU usage on a modern device.
Which devices and browsers does it work on?
It works on modern desktop and mobile browsers that support GPU/WebGL. The model is loaded on first use; after that, it becomes instantly available.