← CHI 2026 map

CHI '26 · Honorable mention · full-paper review · confidence medium-high

VueBuds: Visual Intelligence with Wireless Earbuds

Maruchi Kim , Rasya Fawwaz , Zhi Yang Lim , Brinda Moudgalya , Hexi Wang , Yuanhao Zeng , Shyamnath Gollakota

DOI PDF Program page Page JSON

VueBuds is a strong CHI systems paper because it turns an apparently implausible wearable category into a workable visual-intelligence platform and backs that claim with a substantial evaluation. The novelty is mainly architectural and empirical: the paper shows that earbud-mounted cameras plus host-side VLMs can be viable, while also making clear where the hardware breaks down.

Axes Lens

Rare contribution shape, typical evidence profile. The point here is not a score. It is to show what kind of claim the paper makes, and whether the evidence pattern is unusual or baseline in this 268 -review set.

Review Summary

VueBuds stands out because it challenges a very intuitive boundary in wearable computing: earbuds are normally treated as audio-only devices, and the paper makes a credible case that this assumption is too restrictive. The contribution is not just a gadget demo; it is a system architecture that combines low-power cameras, Bluetooth transport, and on-device VLM processing into a coherent end-to-end pipeline for egocentric visual intelligence. That is a meaningful departure from the common-sense view that the earbud form factor cannot support vision because of size, power, and occlusion constraints. The evaluation is also unusually substantial for a prototype in this space: the paper reports online and in-person studies with 90 participants and compares VueBuds against smart glasses across 17 VQA tasks, with performance described as on par with Ray-Ban Meta. That gives the work real empirical weight and makes the feasibility claim more convincing than a simple lab demo would. At the same time, the paper is appropriately bounded by hardware realities. The evidence points to low sensor resolution, limited dynamic range, glare, and pose-dependent coverage as the main failure modes, which means the system is best read as a proof of viability for a constrained design family rather than a universal solution for wearable vision. In CHI terms, the paper’s value is in opening a new design space and showing that a previously implausible platform can support useful visual intelligence under tight constraints, while being honest about the limits of the prototype and the interaction conditions under which it works best.

What Changed

Canon before

Wireless earbuds are typically treated as audio-first wearables because their small size, battery budget, and radio constraints make embedded sensing and vision seem impractical.

Departure from common sense

Wireless earbuds are usually assumed to be too small and power-limited for camera-based egocentric vision, yet the paper argues that a dual-ear, low-power camera setup can still support useful visual intelligence within those constraints.

Actual novelty

The paper introduces VueBuds as a camera-integrated wireless earbud system for egocentric vision, pairing low-power cameras with Bluetooth streaming and on-device VLM processing. The novelty is not a new vision model, but a wearable system architecture that makes a previously implausible form factor workable for visual intelligence. It combines dual ear-mounted cameras, BLE transport, host-side VLM inference, and opportunistic stitching into a coherent pipeline, then validates that pipeline with both geometric analysis and user studies.

Evidence

The paper supports its core claim with a substantial mixed-methods evaluation: hardware/system design details, geometric blind-spot modeling, VLM benchmarking across five models and two resolutions, and user studies with 90 participants. The evidence shows that the earbud platform can stream low-resolution binocular imagery, that Qwen2.5-VL is the best-performing model in their setup, and that response quality can be comparable to Ray-Ban Meta on the study tasks. The main caveat is that the system is still bounded by low-resolution grayscale imaging, pose-dependent coverage, and desktop-class inference.

Limits

Method limits

Validation is centered on one prototype, one earbud model, and a bounded task set. The studies support feasibility and comparative performance on VQA-style tasks, but they do not establish robustness across all visual-intelligence workloads, all users, or long-term everyday deployment.

Deployment limits

Deployment depends on modified Sony WF-1000XM3 hardware, low-resolution monochrome cameras, Bluetooth streaming, and paired host computation. That makes the system sensitive to battery budget, device compatibility, wearing position, and the availability of a nearby compute device for VLM inference.

Boundary conditions

The paper itself identifies important boundary conditions: low sensor resolution, limited dynamic range, glare, and pose-dependent blind spots can degrade performance; downward-looking scenes without head movement may fall outside the frame; and stitching can fail when parallax exceeds homography-based assumptions.

Manual check: none

Position in field

This is a wearable-systems paper that expands the visual-computing design space into an unusually constrained form factor. Its contribution is primarily architectural and empirical: it shows that camera-equipped earbuds can be a viable platform for visual intelligence, while also documenting the interaction and hardware limits of that choice.

Abstract

Despite their ubiquity, wireless earbuds remain audio-centric due to size and power constraints. We present VueBuds, the first camera-integrated wireless earbuds for egocentric vision, capable of operating within stringent power and form-factor limits. Each VueBud embeds a camera into a Sony WF-1000XM3 to stream visual data over Bluetooth to a host device for on-device vision language model (VLM) processing. We show analytically and empirically that while each camera's field of view is partially occluded by the face, the combined binocular perspective provides comprehensive forward coverage. By integrating VueBuds with VLMs, we build an end-to-end system for real-time scene understanding, translation, visual reasoning, and text reading; all from low-resolution monochrome cameras drawing under 5mW through on-demand activation. Through online and in-person user studies with 90 participants, we compare VueBuds against smart glasses across 17 visual question-answering tasks, and show that our system achieves response quality on par with Ray-Ban Meta. Our work establishes low-power camera-equipped earbuds as a compelling platform for visual intelligence, bringing rapidly advancing VLM capabilities to one of the most ubiquitous wearable form factors.