Object Recognition

Visidon’s object recognition technology is designed to rapidly detect pre-trained objects from images and camera view finder. Objects can be any kinds of natural or human made objects, such as logos, book covers, or buildings, for example. Recognition is done by quickly comprising the object into visual representation defining this specific target with a unique way and comparing it against the database of pre-trained object models.

Technology is highly optimized for mobile devices and is easy to integrate for digital cameras, mobile phones, web cameras, and other camera equipment devices.

Key Features

  • Simultaneous recognition and location of multiple objects
  • Works with any previously trained image object
  • Invariant to translation and scale (up to 8x zoom), and rotation up to 30 degrees
  • Only partial visibility required
  • Compatible with various cameras and input formats; no color information required