chege-unix/wona

An open-source assistive vision system for blind and low-vision users in Africa. It runs on affordable consumer smart glasses (Meta Ray-Ban, Xiaomi and compatible devices) and delivers real-time (smart:usable ) audio descriptions of the surrounding environment — so users can hear what they cannot see. Wona means "to see" in Chichewa and Nyanja, spoken across Malawi, Zambia, and Zimbabwe. Wona is an open-source assistive vision system for blind and low-vision users in Africa. It runs on affordable consumer smart glasses — starting with Meta Ray-Ban — and delivers real-time audio descriptions of the surrounding environment, so users can hear what they cannot see. What It Does Wona listens through the camera on your glasses and speaks back to you. But it does not describe everything it sees — that would be overwhelming and ultimately useless. The hardest problem in assistive vision is not detection. It is knowing what is worth saying. A camera sees hundreds of things in every frame. Describing all of them creates noise that drowns out what actually matters. A person navigating a busy street does not need to hear about the sky, the wall, or the parked car twenty metres away. They need to know about the step directly in front of them, the matatu pulling out, or the shop they are looking for. Wona is built around the principle of intelligent silence — only speaking when something is relevant, urgent, or requested. It prioritises by proximity, motion, and context. It learns what you are doing and adjusts what it tells you. The result is a system that feels like a calm, attentive guide rather than a machine reading the world aloud. What Wona communicates, and when: Obstacles and hazards — flagged immediately, with direction and distance Text and signage — read on demand or when you are clearly looking for information Currency — identified when you are holding a note or coin Faces and people — acknowledged when socially relevant, not constantly narrated Location and landmarks — surfaced when navigating, not when standing still African language output — delivered in the language you choose Design Philosophy Most assistive vision systems treat output as a firehose — detect everything, describe everything, let the user filter. We believe that is the wrong model. Constant …