Skip to content
Karmyug

Data for interfaces with no keyboard, where the input is a hand, a glance or a spoken command.

Hand tracking, gaze, spatial audio and gesture data for headsets.

Spatial computing removes every affordance a model can rely on. There is no click, no cursor and no consistent lighting, so the training set has to carry the variation instead.

We collect gesture, gaze and spatial audio in real rooms with real clutter, across body types and lighting conditions that a lab capture never produces.

Where AI is being applied in ar / vr

The applications teams actually come to us for, rather than the full theoretical list.

  • Hand and gesture tracking

    Keypoint annotation across skin tones, jewellery and occlusion.

  • Gaze estimation

    Eye tracking data with calibration metadata and headset pose.

  • Spatial audio

    Multi-channel capture with room impulse and source position labelled.

  • Scene understanding

    Segmentation of real interiors for occlusion and plane detection.

  • Avatar and expression

    Facial expression capture for driven avatars.

  • Voice commands

    Wake words and short commands recorded in-headset.

Data security and compliance

The controls your risk team will ask about, before they ask.

  • GDPR

    EU data subject rights and DPAs

  • CCPA

    California consumer privacy

  • DPDP Act 2023

    India data protection, in-region storage

AR / VR questions

Can you collect inside a specific headset?

Yes, where we can source the device. Device model and firmware are recorded per session.

Do you handle biometric consent?

Face and gaze data runs under a separate consent template that names the biometric use explicitly.

Power your ar / vr AI with data you can audit.

Start with a pilot batch against your real guidelines. You accept it before anything scales.