[navigation]
Key takeaways
- Banuba Face AR SDK processes video on-device, so patient and clinician frames never leave the app for a third-party cloud and AR processing stays inside the provider's own privacy perimeter.
- Banuba's virtual backgrounds and background replacement run at a stable 30 FPS on a 1280x720 camera, so a real room stays hidden without the video stuttering during a consultation.
- The video-call build adds a 12 Mb minimum footprint, which matters most on the web, where it affects page-load speed and bandwidth for patients on weaker connections.
- Banuba's AR layer sits on top of the existing call transport (Agora, Amazon IVS, OpenTok), so a team adds effects to its current stack instead of rebuilding the call itself.
- One integration covers iOS, Android, Web, Flutter, and React Native, so the same background and touch-up features ship across a provider's patient and clinician apps.
Why do telehealth providers need AR effects on video calls?
A video visit exposes two private spaces at once: the patient's home and the clinician's setting. Virtual backgrounds replace both with a neutral, professional scene, which protects privacy and keeps the consultation focused. Subtle touch-up and skin evening reduce the camera anxiety that makes some patients avoid video care, while background blur keeps family members or a busy clinic out of frame. The result is a calmer, more professional call that patients are more willing to keep and complete.
How does a video conferencing API add these effects?
A video conferencing API like Banuba's supplies the whole AR layer as a single integration. Banuba's Face AR SDK detects and tracks the face and body, separates the person from the background, and then replaces everything else with a still image, a GIF, or a video. On top of that, it applies touch-up, color filters, and 3D masks, and it supports multi-face tracking for group visits, limited only by the device's performance. Because the SDK is convolutional-neural-network-based and runs on-device, none of this depends on a cloud round trip: Banuba's virtual backgrounds hold a stable 30 FPS on a mid-range device, and the video-call build stays at a 12 Mb minimum footprint so web patients are not paying for a heavy download.
What does the integration take?
Banuba's Face AR SDK is a visual layer that sits on top of the call transport a telehealth app already uses, so the call stack itself does not change. Banuba distributes it as CocoaPods and Maven packages and provides platform samples and a video call integration guide. Teams start from the Banuba videocall documentation and the ready-made sample projects, for example the web video-call sample, then drop their client token into the player config and serve the experience. The same approach works across iOS, Android, Flutter, React Native, and desktop builds.
Is it proven in production?
Yes. The same Banuba Face AR layer already powers real deployments, including a live-streaming app built with Amazon IVS and Banuba Face AR SDK, documented step by step here. Because the AR effects run identically whether the transport is a broadcast pipeline or a one-to-one video visit, a telehealth team is adopting a background-replacement and touch-up layer that has already shipped in production video.
Get started
If you are adding virtual backgrounds, touch-up, and AR effects to patient visits, Banuba's video conferencing API gives telehealth teams an on-device, cross-platform layer with a free trial to test against a real call.
-
Banuba's Face AR SDK processes video on-device with no cloud dependency, so patient frames stay inside the app's own infrastructure rather than passing through Banuba. Overall HIPAA compliance still depends on how the telehealth provider handles transport, storage, and hosting, but keeping the AR layer on-device removes one common source of exposure. You can review how the on-device Banuba Face AR SDK handles processing before you build.
-
Banuba's Face AR SDK covers iOS, Android, Web, Flutter, React Native, and desktop (Windows, macOS, Ubuntu) from one integration, so a provider ships the same effects to patient and clinician apps. The Banuba videocall documentation lists the exact camera and OS requirements per platform.
-
No. Banuba's background replacement and face tracking run locally on the user's device, so the video feed is not uploaded to Banuba for processing. That on-device model is what lets the virtual background hold a stable 30 FPS without a network round trip.
-
Yes. Banuba's Face AR SDK supports multi-face tracking in a live feed, applying filters and backgrounds to every detected face, limited only by the device's performance. The same conferencing layer is used in video-calling products such as the VROOM app.
-
Because Banuba's AR layer plugs into the transport an app already runs, most teams integrate from the ready-made video-call samples rather than building computer vision from scratch, which is what shortens the work from months to a focused integration.
