face tracking hero
3D Face Tracking Software

Detect, track and augment faces in a front-camera with our face tracking software for Android, iOS, Windows, Unity and Web.

How Our Face Tracking Differs

Most face tracking solutions need significant computing power. They impose limitations to Face AR experiences, especially on mobile devices with limited CPU. To solve that, we trained the technology to "see" the face in 3D.

  • ft-steps-01
    Live face tracking in 3D

    Unlike face landmark SDKs, our facial tracking technology builds on the 3D math model. It tracks the face by identifying 37 characteristics represented as face morphs. They include different facial expressions, anthropometry and face position in the video frame. It allows us to accurately track the face in the camera and optimize our face detection software for mobile platforms.

  • ft-steps-02
    Mobile-first

    We optimize our technology for mobile devices. It functions well on iPhone-Xs and $100-Androids, at low lighting, and with up to 70% occlusion. Our device support starts with iPhone 5S, Android 5.0 with Camera 2 API and OpenGL ES 3.0. It covers 97% of all iOS devices and 80% of all Android ones.

  • ft-steps-03
    Made for Face AR

    Our Face Tracking SDK precisely detects and tracks several faces in a video stream in real-time. We develop it for Face AR, enabling you to create the most realistic and feature-rich facial animation. You can overlay 3D objects on the face, ad AR lens or turn faces into live avatars and emojis.

  • Performance
    • Multi-face tracking
    • Distance up to 7 meters
    • High performance (min 25-30 fps on early devices)
    • Extreme angles, ranging from –90 to +90
    • Operates in poor lighting
    • Supports up to 70% face occlusion
    • Stable detection with glasses and complex haircuts
    • A 3D model of a face with up to 3,308 vertices
    • Supports 360 degrees rotation of the camera

How We Achieve Industry-Best Face Tracking

Our face capture software runs anywhere - on mobile, web and desktop. It adapts to the device capacity and delivers exceptional performance of Face AR experiences.

  • hw-ach-patented

    Patented anti-jitter

    We run algorithms several times within one frame to process the noises. It results in jitter-free user experience without lag.

  • hw-ach-adaptive

    Adaptive

    Our face tracking supports low lighting, 360-degree device rotation and up to 70% occlusion.

  • hw-ach-efficient

    Efficient energy use

    Less battery drain due to the saved capacity makes your apps lightweight and energy-efficient.

  • whitelabel-about01

    Runs anywhere

    You're not limited to ARCore or ARKit. We optimize our face tracking for low-end mobile devices allowing you to maximize your audience reach.

  • hw-ach-customization

    Customization

    We can adapt our face detection and tracking algorithms to custom hardware.

  • hw-ach-robust

    Robust

    We use compressed neural networks trained on balanced datasets. It ensures our technology works well for all types of skin color, ages, genders and nationalities.

Application Fields

Face tracking can serve as a key component for Face AR apps or as the first step of any face-related computer vision task. It includes facial features classification, emotion recognition or liveness detection.

  • facetracking-fields-01

    Face filtering

    Immerse your users with Snapchat-like face filters, lenses, 3D masks for any camera app.

  • facetracking-fields-02

    Try-on

    Boost sales and advertise new products with realistic virtual try-on tools for facewear.

  • facetracking-fields-03

    AR beauty

    Add virtual makeup testing to help your consumers innovatively experience beauty products.

  • facetracking-fields-04

    Face modification

    Empower the camera with face touch up tools that slim down the contour, emphasize eyes, smooth skin, whiten teeth.

  • facetracking-fields-05

    Photo / video editing

    Allow users to enhance videos and photos with Face AR effects and boost engagement and content sharing in your app.

  • facetracking-fields-05

    Games

    Create the shared in-game experience where players can see each other, overlay face filters or avatars.

Technical Features & System Requirements

  • logo-apple

    iOS

    iOS 10.0+ iPhone 5s and later

  • logo-android

    Android

    Android 5.1 (API level 21)

  • logo-unity

    Unity

    Any mobile platform: ARMv8

  • logo-windows

    Windows

    Windows 10+

  • Web

    WebAssembly support required

  • ft-why
    Why Banuba’s Face Detection SDK
    • 100% proprietary technology and published patents
    • Guaranteed data privacy and compliance with GDPR due to offline work
    • Technology development and optimization for client’s hardware
    • Own R&D center specifically focused on Face AR
    • 7-year team expertise in AR content development and design
    • Trusted by global brands, enterprise partners like Gucci, Sberbank, Yandex
    • Fast support and team member consultancy
    • Hands-on experience in face tracking app development and release

FAQ

  • Our face tracking technology is not built on landmarks. Instead of landmarks, we recognize 37 face characteristics represented as morphs for the default face mesh. This is the most optimal number that allows us to get all the needed data to detect and track a face in the camera.

    When demoing our technology, you may see 87 landmarks marked by which you can evaluate the accuracy of our face tracking.

    • Since we don’t track landmarks but recognize only 37 face parameters, we achieve a much faster technology performance and better accuracy. The algorithms need to process only 37 numbers saving the capacity resources.
    • Patented anti-jitter. As we significantly save on the capacity resources we are able to run algorithms several times within one frame. It shows us the constant data (face itself) and the variable one (the noise). We detect and process the noises within one frame due to which the technology provides no lag.
    • Our face tracking technology is optimized specifically for mobile to function equally well on iPhone-Xs and $100-Androids, at low lighting, and with up to 70% occlusion.
    • Our device support starts with iPhone 5S, Android 5.0 with Camera 2 API and Open GL ES 3.0 which covers 97% of all iOS devices and 80% of all Android ones.
  • Yes, we use the balanced datasets that include people's faces of different ages, genders and nationalities.

Contact Us

Contact Us

Email us for consultancy or technology inquiries.