Version 2022.2
Release Date: 08.06.2022
Update size (data consumption):
  • P100/OP100:
  • P101/OP101 & NVIDIA jetson devices:
Fresh installation size (data consumption):
  • P100/OP100:
  • P101/OP101 & NVIDIA jetson devices:

NEW FEATURES

Rule Engine

Get from Big Data to relevant data - by customizing your configuration by defining rules on given occurrences.
The custom Rule Engine is designed for your individual setup, where data can be filtered already on the edge. On top, rules can be created to get additional information out of your scene by defining occurring situations by a rule.
The defined rule will send a specific event if it occurs. Special events as for example, wrong-way driving, unexpected turnings, unexpected class(-es) in dedicated areas and much more can be defined based on the existing event triggers.
See the options and how to use the rule engine in detail further back in the documentation. The following video gives an overview about one use case showing a safety issue while a pedestrian is crossing the street.
Pedestrian crossing safety issues
In Data Analytics, you can visualize how often the events occurred in a given time frame with the standard feature set of Data Analytics.
Rule Trigger Count
If you would like to get the information in close to real-time, you can use the MQTT data integration option, where all the details of the event will be pushed to the broker. You can find the new event schema and all respective details in the following section of the documentation.

ANPR Enhancements & Country Meta Information for ANPR

During this launch, we have extensively been working on ANPR in order to improve its accuracy. Therefore, we could extend a solid ANPR solution up to 40 km/h speed with an accuracy of 90% on a challenging outdoor scene at entry/exits without any special slow down installations. In addition, we have added more meta-information on the license plate detection. It will be possible to see the issuer country of the license plate. This will allow you to get more information about your parking visitors for use cases like: payment solutions, enforcement, or for identifying the region where the parking customers come from.
The country codes are according to ISO 3166 Alpha 2 standard. The country classification is working with excellent accuracy of 99%.

Further Improvements

Traffic model improvements with a focus on vulnerable road users and stability

With this launch, we have as well worked on an even better detection with our Traffic & Parking Model in order to have more stable accuracy figures within different environmental conditions. We were mainly focusing on excelling in sunny, bright conditions where shadows could cause issues. On top of that, pedestrian and bicycle detection has been improved significantly.
Receive the license plate capture via MQTT
With this release, we provide a special option to have the license plate capture added to the event schema. This enables you to retrieve the capture within the pushed MQTT message. The picture is encoded in BASE64. In order to enable this option, please contact our support.