VeMIR (Visual Engines Mobile Image Recognition) is a software development kit (SDK) to add image recognition capabilities to mobile applications developed for both iOS and Android. By embedding VeMIR it becomes possible to activate the cam of the device, to scan the surrounding reality to find objects or images. Once the object is recognized, it can be described in detail and used inside the application itself.
|Would you like to try with your own images?|
how it works
BEHIND THE SCENE
The first step is to upload your images on the VeMIR content repository called “VeMIR Recognizer Builder“. The repository is a Web server where all your images are stored, enriched with useful descriptive information and processed to extract the specific signatures. Creation of search indexes on image signatures is the crucial step enabling the matching process.
VeMIR Recognizer Builder Turorial
SYNCHRONIZE MOBILE DEVICES
All the mobile devices containing a VeMIR enabled application are periodically updated with data from the VeMIR Web Server. In particular, the signature database and the thumbnail repository on the device are updated with respect to the new images acquired on the server.
SCAN THE WORLD AROUND YOU …
Open your VeMIR enabled App and start scan the real world around you with the cam. Point the cam toward objects you want to know more about i.e. a painting on the wall, the front of a church along the street, an image on a magazine. The scan process may be applied on single shoots (the user point the cam and click to take the picture to be matched) or on a continuous scanning (the user open the cam and move it around, VeMIR automatically take the pictures and match them).
FAST AND WORKING EVERYWHERE
The VeMIR App may work totally off-line matching all images taken by the cam with the local (device) signature database. The “on-device” nature of the matching mechanism make it possible to use it everywhere, even on an airplane, abroad or in places where connection isn’t available. Moreover, the matching algorithm, if not using the network, is extremely fast as it takes few milliseconds to match a single image.
The probability to recognize an object, the image of which is stored in the VeMIR repository, is very high. The matching algorithm is able to cope with various deviation from the exact image present in the database (rotation, translation, immersion in a different context, etc.).
When the object is know (recognized), the application will tell you more about it.