Purpose of the solution

The functional characteristics of COS.TRACKER provide solutions to the following tasks based on technical means of video surveillance infrastructure:

• viewing the current state of the complex, management and configuration; 
• control the operation of plugins on the node;
• building a node architecture and managing plugin parameters;
• database access control; detection and configuration of IP cameras;
• capturing video frames from machine vision cameras, incl. Hikrobot, Daheng, Basler;
• buffering of input video frames for frame-by-frame synchronization in video analytics; video stream capture;
• hardware encoding/decoding of video stream; broadcasting a video stream to the network;
• detection of failures in the operation of the IP camera;
• motion detection in a video stream; searching for objects in the image frame; data capture from lidars, incl. manufactured by SICK, LS Lidar, G-TEK, Cyndar;
• processing of lidar readings; formation of 3D images of objects; allocation of vehicles;
• determination of state registration plates of vehicles and generation of metadata based on the results of their recognition;
• vehicle classification;
• formation, viewing and transmission of transactions; work with various NPUs, incl. Rockchip, Sophon, Jetson.

   

Sources of information

The following can be used as sources of information:

• high-resolution cameras with USB/GbEth/5GbEth connection, incl. computer vision cameras Hikrobot, Daheng, Basler; lidars SICK, LS Lidar, G-TEK, Cyndar;
• other sensors, information from which is aggregated in databases managed by PostgreSQL / SQLite DBMS

Documentation

Documentation containing a description of the functional characteristics of the software and information necessary to install and operate the software is available at the links below: 

• Installation information for COS.TRACKER >>>>>
• Description of the COS.TRACKER web user interface >>>>

Pricing

The cost of the software “Program for tracking objects in machine vision systems COS.TRACKER” depends on the set of business problems being solved and the terms of the license agreement and is calculated individually. The terms of the license agreement can be either a simple non-exclusive license or a complete transfer of exclusive rights. To receive a commercial offer, you must send a corresponding request to info@cos.ru.

Применение нейросетевых технологий для автоматической классификации и маршрутизации текстовых сообщений и документов в работе ЕЦОР и служб эксплуатации и ремонта

Хельвас А.В., Овчинников А.Д., Кузнецова А.А., Гиля-Зетинов А.А.

в сборнике ”Имитационное и комплексное моделирование морской техникии морских транспортных систем” Материалы пятой международной научно - практической конференции ИКМ МТМТС-2019, место издания М: ”Издательство Перо”, с. 191-196

The software platform and information model of a situational center

S.S.Shcherbakov, A.A.Kuznetsova, A.A.Bespalko, A.S.Galitskiy,  A.V.Khelvas

Trudy MFTI 2018 v.10 №4(40) pp.98-109

The article deals with message standards for transmitting information on emergencies and a software platform for sharing them between government agencies and the full range of organizations related to emergencies.

Practical usage of Alloy for tactical situations recognition

A.V. Khelvas, A.B. Galitskiy

Trudy MFTI 2018 v.10 №4(40) pp.87-97

The article discusses the approach to the detection of tactical situations based on the analysis of formal statements generated by processing the flow of messages about events and threats in a formalized form in the software platform of the operation center . The analysis uses the Alloy language and the Alloy Analyzer analytical application. An example of an analysis for the typical flood threat detection situation is also discussed.

Improved 2D Human Pose Tracking Using Optical Flow Analysis

Khelvas Aleksander, Gilya-Zetinov Alexander, Konyagin Egor, Demyanova Darya, Sorokin Pavel, Khafizov Roman in book Advances in Intelligent Systems and Computing, series AISC, Springer, v. 1100, pp. 10-22   DOI: 10.1007/978-3-030-55187-2_2 In this paper, we propose a novel human body pose refinement method that relies on an existing single-frame pose detector and uses an optical flow algorithm in order to increase quality of output trajectories. First, a pose estimation algorithm such as OpenPose is applied and the error of keypoint position measurement is calculated. Then, the velocity of each keypoint in frame coordinate space is estimated by an optical flow algorithm, and results are merged through a Kalman filter. The resulting trajectories for a set of experimental videos were calculated and evaluated by metrics, which showed a positive impact of optical flow velocity estimations. Our algorithm may be used as a preliminary step to further joint trajectory processing, such as action recognition.