Tracking of moving objects is a part of many applied tasks in computer vision. The first step in dealing with such tasks is to identify the objects in the original video clip, in other words, to locate the regions of interest (ROI), that will be further employed to ensure the tracking. This paper describes an algorithm and a cross-platform software that allows to analyse algorithms written in various programming languages based on the YouTube-BB dataset. The paper provides examples of our systems work based on the algorithms available in the OpenCV library and describes an interaction protocol allowing to test alternative algorithms.