Dallmeier looks at how video technology is getting smarter

El fabricante alemán Dallmeier detalla cómo la videovigilancia, la analítica de vídeo, especialmente basado en IA, cobra cada vez más importancia en su aplicación en empresas de todo tipo y sectores.

Las tecnologías de videoseguridad se están alejando cada vez más de la ‘vigilancia clásica’ acercándose a los sistemas inteligentes, subrayan desde Dallmeier. A través de un análisis de vídeo eficiente y el uso de la IA, las cámaras y los sistemas de gestión ya no se limitan a proporcionar imágenes, sino que apoyan activamente al usuario en su trabajo.

En la videovigilancia, la analítica de vídeo va cobrando más y más importancia. The technology, particularmente en el AI-based video analytics, se está desarrollando rápidamente: in addition to the preservation of optical evidence, users have more options for automatic or semi-automatic analysis of image data thanks to this technology.

The man 'brakes'

An analysis of events in video surveillance can be carried out live, For example, when public places are monitored or, in retrospect, in case they have to be limited to the events surrounding a crime.

Traditionally, the evaluation and assessment of video images was and still is performed by humans, that is to say, by an operator in a surveillance center. As Dallmeier points out, the challenge here is that when a person analyzes a video they can only do so at normal speed (not in fast-forward mode). Analizar varios vídeos en paralelo también se antoja muy difícil.

La IA ‘asiste’

Si se van a analizar las grabaciones de docenas o cientos de cámaras, posiblemente de varios días o incluso de semanas, los procesos puramente ‘humanos’ alcanzan rápidamente sus límites de rendimiento, tanto en términos de recursos de personal disponibles como, of course, de costes.

Nowadays, existe en la analítica de vídeo un software que apoya activamente a los investigadores y a los servicios de emergencia. En este proceso se está empleando cada vez más la tecnología de inteligencia artificial.

En principio, las cámaras son muy adecuadas como ‘sensores ópticos’ para captar tales datos de análisis: there are only better ways to extract from complex contexts, with relatively little effort, a wide variety of data than what a video image offers. The recordings, as evidence in court, For example, are included.

The possibilities of video analysis with AI support are varied: the 'crowd analysis' to count people or objects, For example, to ensure safety at large events or to open another counter at the airport if there are too many passengers waiting.

The 'ship detection' allows the recognition of small vessels for the prevention of terrorist attacks or security at ports and oil refinery ports.

The so-called 'smart search' techniques they help to find people or objects based on certain characteristics, such as the color of outer clothing, and various 'intrusion detection' systems improve property protection by distinguishing between false alarms and real threats.

Perimeter protection

the monitoring of narrow strips along perimeters, such as fences, buildings or streets, often called 'sterile zones', is one of the most demanding requirements for physical security.

Until now, los sistemas de vídeo, particularly leaders in high-security environments, often did not meet user expectations: unsatisfactory detection performance, many false alarms and, above all, easy manipulation to evade analysis are frequently mentioned drawbacks.

Among other shortcomings are low image quality and high infrastructure costs, especially due to the large number of video devices required. The new camera system Panomera S4 Perimeter de Dallmeier, solves this problem.

Specifically, Panomera S4 Perimeter is equipped with a neural network specifically trained for perimeter protection and reliably detects people with unusual body postures (bent over, crouching or lying down), and movement patterns (very slowly, very quickly, crawling).

It also reliably detects people wearing camouflage clothing, such as Camouflage Central Europe, British Smock or Ghillie. The advanced AI application for sabotage detection supports the recognition of manipulation attempts that are typically expected at a perimeter.

Alongside classical methods such as turning, blurring, spraying and covering the camera or IR lighting, it also reliably detects device dazzling with a laser pointer, a flashlight or a strobe light.

Additionally, indirect manipulations are detected and alerted through deliberate nebulization of the detection area using pyrotechnics, such as smoke grenades or smoke canisters. The combination of both assessment techniques provides optimal detection of intrusion attempts with a very low rate of errors and false alarms.

Practical case: CTH Herne

Ubicada en Alemania, Container Terminal Herne (CTH) ha apostado en sus instalaciones por una combinación de las patentadas cámaras de sensores multifocales Panomera y un análisis de vídeo basado en IA para proteger una superficie de unos 80.000 m² (more data in Digital Security).

El análisis de vídeo basado en IA actúa como sistema de asistencia; For example, sólo transmite las alarmas verificadas a una central receptora de alarmas externa. This way, los operadores tienen menos carga, y la protección mejorada contra robo y la cobertura incrementada de seguro hacen que el emplazamiento sea más atractivo para clientes existentes y nuevos.