Visual Objects Identification System

Use:

"Visual Objects Identification System" acts as a pre-filter and is responsible for the identification of visual media data already known to the system, e.g. images or video.

Functionality:

• Dynamic system training, i.e. optimization aiming at preventing decrease in efficiency and recognition accuracy;
• Optimal operation with large database of visual objects. On a modern workstation identification time over a one million image database is not more than a few milliseconds. Use of such large databases in the identification is a unique and unparalleled operation. The difference in speed between the mathematical orders of the database sizes has a logarithmic dependence;
• Low resource intensity and scalability that provide high speed and capacity to link clusters;
• Unique innovative Smilart technologies.

System testing:

The system stability was checked using several millions of images and hundreds of movies. Youtube was successfully examined with the system that found some copyrighted content.