CRITIS security on land and at sea

As part of the critical infrastructure, ports and refineries have a particularly high protection requirement: As, for example, highly flammable and explosive petroleum products are handled and processed, unauthorised persons and vehicles must not be allowed access under any circumstances. In many cases, access from the water side must also be secured in accordance with the ISPS Code (International Ship and Port Facility Security). Intelligent video security solutions can assist in the visual surveillance of these sensitive areas.

The reasons for preventing unauthorised access to ports and refineries are many and varied, ranging from “simple” vandalism to general security and even terrorist attacks. Regardless of the specific reason, unauthorised persons and vehicles must not be allowed to enter the premises, either through gates and fences (on the land side) or through port locks or shore areas (on the water side). In practice, there is another challenge: port operators and refineries also want to prevent the entry of small watercraft. On the one hand, this is to prevent terrorism and, on the other hand, to avert danger, as there is a risk of sparking from outboard engines, e.g., from the ignition system, which could ignite gases and cause an explosion. It is particularly important for operators to comply with the security requirements of the ISPS Code and to maintain complete documentation of each entry of persons and vehicles.

Watercraft challenge: AI helps against light reflections
In the past, operators often used analogue surveillance cameras to secure factory gates and landside parts of a site. However, the ageing technology no longer provides the image quality required to clearly identify people and vehicles. Video content analysis (VCA) cameras are sometimes used to monitor ship locks to detect unauthorised vessels. However, these systems generate many false alarms due to wave action and water reflections. This places a huge burden on security officers to manually verify this high number of false alarms.

There is no standard solution
In video technology, the problem of high false alarm rates is typically associated with perimeter situations where people or vehicles are being detected. However, the challenge in ports and refineries is to detect objects automatically and reliably despite waves and water reflections. This requires large amounts of video to train neural networks on these waterborne vehicles.

For example, ports and refineries are now using video security solutions that monitor entire locks and automatically detect small watercraft with these specially trained neural networks. But since any video analytics is only as good as the data it has, AI video analytics is only as good as the images and video it receives from the cameras.