Text Periscope

The future of text detection: real time recognition and playback of letters, words and sentences filtered from natural environments.

Thanks to powerful smartphones and tablets, instant text detection and output on the go become possible. Our Demo TextPeriscope already shows what is possible today.

Here is how it works (currently on iOS):

  • the camera of the smartphone or tablet is pointed at any text in a natural scene (can be a document, a sign, a product description, etc.)
  • TextPeriscope scans the video stream for text, once it detects text with reasonable quality in a pre-defined area of the video stream (e.g. the center of the screen), it extracts the text snippet
  • this text snippet is sent to the OCR engine which translates the snippet into characters
  • finally the characters are then spoken out via a Text-To-Speech engine

And all steps are done in real time!

***video demo coming soon***

What’s next

We want to build a tool that explores and explains the world around the user. Know that there is text is the first step that we have tackled with our Text Detection algorithm. Now the question we’d like to answer is: where exactly is the text in space, how we can effectively communicate the spatial postion to the user, and which text is relevant to the user, based on context, conventions and preferences?

With new user interface concepts, advanced image and text processing techniques and more accurate OCR technology, many areas exist that can be explored to make reading text from natural environments fast, flexible and accurate

Will it remove the final barriers for people with visual difficulties to find and read text? – we’re excited to work it, test it and eventually realize this vision.