Mobile Visual Search

Server-free large scale image recognition, especially for mobile devics such as smartphones and tablets.

  • Full control over your data: Trainable on your standard desktop
  • Scalability: Local recognition of more than 10 000 images
  • Extensibility: Support for learning new images at runtime
  • Low latency: Local processing instead of network transfers
  • Always available: No dependency on network status
  • Ready for Augmented Reality: Tracking by detection for initialization of Natural Feature Trackers

Note: The original approach was developed at Graz University of Technology for Bundesdruckerei GmbH.


  • A. Hartl, D. Schmalstieg and G. Reitmayr. “Client-Side Mobile Visual Search”, Proceedings of the International Conference on Computer Vision Theory and Applications, 2014