AccessEye

AccessEye describes the world to people who cannot see it. Point your phone, tap anywhere on the screen, and you hear what is in front of you, in your language.

I built it because most tools of this kind need an internet connection, an account, and they send your camera feed to a server somewhere. None of that happens here. An AI model runs entirely inside your phone, looks at the photo locally and tells you out loud what it shows: people, obstacles, objects, signs and any text you might need.

The whole screen is the capture button, so there is nothing to hunt for. The app speaks for itself at every step, which means it works comfortably even without TalkBack, and when TalkBack is on the two cooperate properly so you never hear two voices talking over each other. There is haptic feedback for every action, and the interface uses large, high contrast elements.

It speaks nine languages: English, Greek, Spanish, French, German, Arabic, Hindi, Italian and Russian. The same model does both the description and the translation, so changing language is just one setting. Every description is saved with its photo in the history, so you can listen to it again any time.