VisionBuddy

An accessible camera prototype that detects common objects and speaks what is in front of the user.

Camera preview area.

Start Camera Describe What Is Ahead Start Auto Scan Stop Speaking

Loading object detection model...

Spoken Description

No description yet.

Detection confidence: 50% Auto scan interval: 4 seconds

Safety note: This prototype can miss objects or identify them incorrectly. It should support awareness, not replace a cane, guide dog, mobility training, or normal safety judgment.

Technical note: This page requires camera permission and works best from HTTPS or localhost. The object detection model loads from the internet using TensorFlow.js and COCO-SSD.