How it works?

The basis of this service is a complex algorithm for recognizing and cataloging objects in images and video.

The algorithm is based on a self-learning neural network and is currently able to recognize objects in a frame and some actions. The algorithm does NOT understand social roles or occupations. For example: advocate or gamer. Therefore, during the search, please try to provide exactly which objects are present in the frame.

To start a search, enter the name of the object in the search box. For example: phone

You can also expand your search query and write in natural English. For example: phone in hand

If you want to find a specific phrase, the phrase must be written in the quotation. For example: "sports car" In this case, search results will consist of images that contain sports cars. Otherwise, cars will be made up of sports elements and cars.


What objects can be found?

The database stores more than 10,000 different objects and this amount is increasing every day. When you make a request, it is best to imagine not an action in the frame, but what objects are involved in this action. For example: man in car or man driving car

You can also try to look for various abstract queries like: art or pattern