AI-powered smart glasses for visually impaired users in Rwanda
# 🥽 Inzozi Glasses
## Bilingual AI-Powered Smart Glass System for Visually Impaired Users in Rwanda
### Project Overview
Inzozi Glasses (meaning "Dreams" in Kinyarwanda) is a wearable smart system that helps visually impaired users navigate indoors, recognize faces, detect emotions, and receive voice guidance in both Kinyarwanda and English.
### Features
- **Object Detection (YOLOv8n)** — Detects everyday objects in view
- **Face Recognition** (`face_recognition`/dlib) — Recognizes registered people from `known_faces/`
- **Emotion Detection** (DeepFace) — Detects emotions on recognized faces
- **Scene Recognition** — Classifies Indoor / Outdoor
- **Distance Alerts** — Optional Arduino ultrasonic sensor + vibration motor for obstacle warnings
- **Bilingual Voice** — Kinyarwanda + English text-to-speech and (in the standalone app) live language switching
### Live Demo
drive.google.com
### Final Product Demo
drive.google.com
---
## Read this before "How to Use" — there are two runnable systems here, not one
This repo actually contains **two separate ways to run Inzozi Glasses**, plus a Windows `.exe` build of one of them. They are not the same thing, and the old "open `index.html` to see the complete simulation" instruction undersold what's actually here — one of these two is a real, working system, not a simulation.
| | What it is | How you run it | What powers detection |
|---|---|---|---|
| **A. Standalone hardware app** (the full-feature system) | Phone camera feed + optional Arduino distance sensor + face recognition + emotion detection + scene recognition + bilingual voice, shown in a desktop OpenCV window | `python smart_glasses_phone_fixed.py`, **or** double-click `InzoziGlasses.exe` if you have the built package | YOLOv8n, `face_recognition`/dlib, DeepFace — all running locally, no cloud calls |
| **B. Browse …