The railway marshalling station occupies a central place in the technological chain of freight transportation processes, since the speed of processing trains at marshalling yards determines the volume and cost of transportation. Therefore, development of automation and computerization of sorting processes results in growing efficiency of freight transportation in general. The objective of the study is to formalize the problem of cars’ monitoring within the railway marshalling yard and to develop a method for solving it with the use of algorithms of recognizing and positioning of dynamic objects through the intelligent data analysis of streaming video. The article presents a new approach to solution of the problem of monitoring moving units in the hump (sorting) yard of marshalling stations. The article suggests core criteria for identifying speed and positioning of the railway wagons when they are running after been separated at the hump. The article specifies that monitoring of moving units at hump yard is less automated in comparison with the monitoring at the hump itself, and that confirms the relevance of the research. To get the problem of the automation monitoring of moving units in the hump yard solved, the authors have suggested an algorithm that is based on the image data intelligent analysis, that is on computer vision, and have described the model of its implementation at a station. The methods used are based on the theory of computer vision and are aimed at recognizing key dynamic objects in streaming video and at their subsequent positioning. The study has resulted in substantiation of acceptability of the use of computer vision in the process of separation and formation of trains. It is planned to proceed with further improvement of the presented approach to develop a software product allowing to objectify information about hump yard in order to increase the efficiency of targeted braking at the hump.