Voice commands and speech feedback across ten Indian languages
Engineering case studyShobhit Tripathi / 2026
03 / Voice-first Assistive Computer Vision
VisionMate
03
01Voice / camera→
02Navigation→
03Detection / OCR→
04Processing→
05Spoken assistance→
06Caregiver live view
01 Overview
THE SYSTEM,
IN CONTEXT.
VisionMate combines voice control, walking navigation, live object detection, OCR, SOS, and real-time caregiver location tracking across three independently deployable services.
Screen-first navigation creates barriers for visually impaired users, while caregivers lack a continuous, low-friction view of location and emergencies. VisionMate makes voice the primary interface and the camera a contextual aid.
02 System workflow
01Voice / camera
02Navigation
03Detection / OCR
04Processing
05Spoken assistance
06Caregiver live view
03 Key features
Walking routes with spoken turn guidance and saved routes
YOLOv8n object detection with position, risk, and approximate distance
On-demand EasyOCR text reading
Socket.IO caregiver tracking and live SOS alerts
JWT role-based accounts and a secured administration portal
04 Engineering
BUILT AS
A WORKING SYSTEM.
The implementation is presented from its verified live behavior and public source. No usage metrics or unverified performance claims are included.
React 19ViteNode.jsExpressMongoDBSocket.IOPythonFastAPIYOLOv8nEasyOCRLeafletWeb Speech API
05 Current state
Working three-service application
The deployed application can be opened from the live demo link.