Seeing AI is an artificial intelligence camera application developed by Microsoft. This app can help visually challenged individuals seek information from their surroundings on their own. You can download the app from the App Store.
What you can do with Seeing AI:
From identifying people and products to reading out documents and signs, Seeing AI can aid anyone with a visual impairment to be independent. To use the app, you just have to point the camera at any object and swipe the screen from one side to another to know different kinds of information. Let us now explore all the features of Microsoft's Seeing AI.
This channel is used to read small amounts of text like signs, addresses on envelopes, etc. Currently, Seeing AI cannot read handwritings.
This channel can scan documents and read it out to you. It does this by telling you if it can see the entire document. If the entire document is not in the frame, the AI announces for example, "Top edges not visible". This then gives you an idea of how to adjust your phone in order to capture the whole document. Once the whole document is in the frame, the AI then captures an image of it and reads it out to you.
With Seeing AI, you can identify products based on their barcode. All you need to do is hold your phone up to the product with the help of audio cues. The app beeps faster when it is close to the barcode and slowly when it's not. When it is close enough to read the barcode, you get an auditory description of what the AI thinks the product is.
This channel can recognize and describe people around you. It even gives you an estimate of the person's age and their expression or emotion. If you want Seeing AI to learn and remember a person's face, all you have to do is take 3 pictures of the person from the Face Recognition tab found in the menu and type in their name. Next time the AI sees a known face, it will tell you who it is.
The scene channel helps you understand what is around you by describing the scene when you take a picture. The Seeing AI scans through the photo and tries to recognize objects and tries to figure out the connection between them. Once it has worked out a connection, it then describes it's understanding through simple language. This channel, however, is an experimental one so it has higher chances for mistakes.
This channel can help you recognize money. Once you hold the camera up to the note, it will tell you the denomination as soon as it recognizes is. Unfortunately, this feature currently works only for US currency.