Video analytics for public transport: Big Data - the underwater part of the iceberg
Intellectualization of video surveillance in transport is one of the most promising areas of the industry due to the large-scale construction of public infrastructure. So, in Moscow alone, it is planned to re-equip 188 existing metro stations, build 64 new underground stations, 31 ground stations on the Small Ring of the railway and 5 light rail lines with a paid trip to the station. Each underground station will contain at least 50 cameras on which situational and biometric video analytics will be optimized for crowded places.
It is important that the introduction of intellectual video surveillance equipment is mandatory at the level of the federal law on transport safety, orders of the Russian Government on approving the Comprehensive Program for ensuring public safety in transport and Mintras orders on approving transport safety requirements for categorized objects ( more on the regulatory framework for transport ) .
')
Situational video analytics can be implemented on the basis of tracking algorithms for ceiling cameras that accompany people “over their heads”. Situation detectors operate on the basis of the resulting trajectory and rules set by the security service. Thus, it can be configured detectors crowds (crowds), movement against the flow, fast movement (running). It is possible to use specialized detectors to automatically detect the facts of people falling on the rails , the appearance of left objects and the pair passage through the turnstile .
Examples of video analytics
Biometric video analytics is based on facial recognition technology based on biometric facial features. In the simplest case, the search for suspected persons on the "black list". Under more complex scenarios that have not yet become widespread in practice, biometrics are integrated with situational video analytics and the person tracking system. For example, “hares” jumping through a turnstile are automatically entered into a “black list” and with repeated violations can be detained and fined.
A promising multi-chamber biometrics, which involves a multiple comparison of persons between different boundaries of geographically distributed observation networks. Such a system makes it possible to analyze not only the movement of suspected persons through the public transport network, but also to collect detailed statistics of passenger traffic for its balanced development (see also multi-chamber tracking ).
Video analytics in transport is increasingly being used outside the security service for collecting statistical data and making management decisions. Such analytical systems are needed to manage passenger traffic , improve the quality of service and staff productivity.
If you look at the analytical systems for transport from the point of view of modern IT trends in general, the “underwater part of the iceberg” of the upcoming work opens. Video analytics and biometrics today most often work at the object level, and the collected metadata with are used only to attract the attention of the operator. The metadata archive, which hides a lot of useful information, is not collected or analyzed.
Among the ten strategic technologies identified by Gartner for 2013, at least three strongly correlate with the tasks of transport analytics: Strategic Big Data, Actionable Analytics, and Mobile Apps. Similar forecasts are made by IMS Research: Metadata as Big Data (Video Meta-data as “Big Data”) and Video Surveillance (Video Surveillance Goes Vertical).
Thus, the full implementation of video analytics in transport involves the use of fundamentally new approaches for storing and analyzing large amounts of metadata. Private tasks here are searching for objects and situations in a distributed video archive, filtering and ranking events , aggregating and visualizing data for reports with their contextual dependencies taken into account.
It is important that the introduction of intellectual video surveillance equipment is mandatory at the level of the federal law on transport safety, orders of the Russian Government on approving the Comprehensive Program for ensuring public safety in transport and Mintras orders on approving transport safety requirements for categorized objects ( more on the regulatory framework for transport ) .
')
Situational video analytics can be implemented on the basis of tracking algorithms for ceiling cameras that accompany people “over their heads”. Situation detectors operate on the basis of the resulting trajectory and rules set by the security service. Thus, it can be configured detectors crowds (crowds), movement against the flow, fast movement (running). It is possible to use specialized detectors to automatically detect the facts of people falling on the rails , the appearance of left objects and the pair passage through the turnstile .
Examples of video analytics
Biometric video analytics is based on facial recognition technology based on biometric facial features. In the simplest case, the search for suspected persons on the "black list". Under more complex scenarios that have not yet become widespread in practice, biometrics are integrated with situational video analytics and the person tracking system. For example, “hares” jumping through a turnstile are automatically entered into a “black list” and with repeated violations can be detained and fined.
A promising multi-chamber biometrics, which involves a multiple comparison of persons between different boundaries of geographically distributed observation networks. Such a system makes it possible to analyze not only the movement of suspected persons through the public transport network, but also to collect detailed statistics of passenger traffic for its balanced development (see also multi-chamber tracking ).
Video analytics in transport is increasingly being used outside the security service for collecting statistical data and making management decisions. Such analytical systems are needed to manage passenger traffic , improve the quality of service and staff productivity.
If you look at the analytical systems for transport from the point of view of modern IT trends in general, the “underwater part of the iceberg” of the upcoming work opens. Video analytics and biometrics today most often work at the object level, and the collected metadata with are used only to attract the attention of the operator. The metadata archive, which hides a lot of useful information, is not collected or analyzed.
Among the ten strategic technologies identified by Gartner for 2013, at least three strongly correlate with the tasks of transport analytics: Strategic Big Data, Actionable Analytics, and Mobile Apps. Similar forecasts are made by IMS Research: Metadata as Big Data (Video Meta-data as “Big Data”) and Video Surveillance (Video Surveillance Goes Vertical).
Thus, the full implementation of video analytics in transport involves the use of fundamentally new approaches for storing and analyzing large amounts of metadata. Private tasks here are searching for objects and situations in a distributed video archive, filtering and ranking events , aggregating and visualizing data for reports with their contextual dependencies taken into account.
Source: https://habr.com/ru/post/172721/
All Articles