Study of methods of segmentation and detection of objects in the image in real time to prevent accidents of Russian Railways
Journal Title: Modern Innovations, Systems and Technologies - Year 2022, Vol 2, Issue 3
Abstract
With the development of the railway industry, the informatization of society and the automation of many technological processes, it becomes possible to create a complexof automatic control, diagnostics and safety of locomotive traffic. One of the most important systems of this complex is the system for detecting objects on the railway tracks, breaks in the railway track and its turns. Such a system can be designed as a locomotive-mounted camera and information processing systems on board each rolling stock, or as a global system that performs remote processing of information from several locomotives. Regardless of the implementation of the system, there is a need to create a block for detecting objects in images coming from cameras. To implement this block, it is necessary to select a railway lane in the image and detect objects in real time. Segmentation methods are used to select a band. The article presents the algorithms of several of them and chooses the most preferred option. The task of detecting objects in a video stream in real time is solved using convolutional neural networks. The article provides brief descriptions of several networks, analyzes the results of the described neural networks, and selects the network that is most suitable for solving the problem
Authors and Affiliations
A. T. Tisetsky, D. I. Kovalev, T. P. Mansurova
Keywords
Related Articles
- EP ID EP708580
- DOI https://doi.org/10.47813/2782-2818-2022-2-3-0101-0116
- Views 124
- Downloads 0
Reducing the construction time of a monolithic frame of an apartment building
Currently, the construction industry is facing many problems that hinder its development. One of the key obstacles is the lack of new management approaches that could allow adapting to the changing conditions of the cons...
Low light combining multiscale deep learning networks and image enhancement algorithm
Aiming at the lack of reference images for low-light enhancement tasks and the problems of color distortion, texture loss, blurred details, and difficulty in obtaining ground-truth images in existing algorithms, this pap...
Application of the ant colony algorithm for solving the job shop problem
The ant colony algorithm was first introduced by M. Dorigo, V. Maniezzo and A. Colorni in 1991. This algorithm is based on the behavior of a real ant colony when searching for food. The main area of application of the al...
Control method for an autonomous group of multi-rotor aircraft
The purpose of this article is to improve the efficiency of control and monitoring of an autonomous group of multi-rotor aircraft by developing a hardware and software complex. The research is aimed at creating an intell...
Software parameters affecting the reliability of telemetry data processing
Creation of any modern complex technical systems, as well as their management, is impossible without complex software. In this case, the most important technical task is to increase its reliability and fault tolerance. A...