---
title: "Vision-Tracked Hand Pose System"
description: "Hand-tracking glove system running on Raspberry Pi hardware, using camera-based marker detection and temporal filtering for real-time pose estimation."
canonical_url: "https://krishshah.dev/kshah00/projects/gloves"
md_url: "https://krishshah.dev/llm/projects/gloves.md"
last_updated: "2026-05-28T19:45:15.566Z"
project_id: "gloves"
---

# Vision-Tracked Hand Pose System

- **Slug / id:** `gloves`
- **Label:** Embedded AI · Computer Vision
- **Mode:** engineer
- **Category:** rnd
- **Status:** wip

## Summary

Hand-tracking glove system running on Raspberry Pi hardware, using camera-based marker detection and temporal filtering for real-time pose estimation.

## Detail

Started in simulation, then moved into a physical prototype running on constrained hardware. The hard part was balancing tracking accuracy, temporal stability, and latency well enough for the system to feel usable instead of just technically correct.

## Metrics

- **Runs on edge hardware:** Raspberry Pi