Version 2022.1
Release date: 02.03.2022
Update size (data consumption):
  • P100/OP100: ~500MB
  • P101/OP101 & NVIDIA jetson devices: ~700MB
Fresh installation size (data consumption):
  • P100/OP100: ~800MB
  • P101/OP101 & NVIDIA jetson devices: ~1000MB


Single- & Multispace parking E2E solution

Easier event type configuration for Parking and Traffic scenarios with more options for your Parking configurations.
From now on, event-type templates are available depending on the use case you want to implement. The templates are divided in two segments Parking and Traffic. Based on the selected template, you have the already pre-configured trigger settings that can be changed if necessary.
Single-space and Multi-space Region of Interest (RoI) are now dedicated parking RoI that only consider vehicles in the detections. All parking RoI are transmitted as a single aggregated event with all necessary information every 10 seconds, reducing post-processing effort and data traffic.
In order to make the calibration and configuration of RoI easier, there is now the option to show the vehicle detection boxes as an overlay at the configuration stream. There it can be seen where the center point of the vehicles is located and it will give you support on how to draw the RoIs.
Enhanced Parking scenario options by improving Data Analytics with Single- and Multi-space parking data:
With the enhancement of our Parking scenarios within Data Analytics, it is now possible to display parking utilizations based on Single or Multi-space configurations. On top of that, you can create multimodal parking scenarios where you can combine entry and exit capacity counting with Single-space counting for dedicated parking lots in your parking area ( e.g.: EV charging spaces)
  • You can view the capacity of your parking space based on all available Single- or Multi-space parking configurations.
  • See the occupancy status of each configured Single- and Multi-space parking space and quickly access this information via our Data Analytics API.
Improved Parking scenario detection and greater coverage of parking spaces per camera
With model enhancements focused on parking scenarios, we have increased detection accuracy to more than 95% for Single- & Multi-space parking scenarios. With a new model for parking that analyzes the camera frame with higher resolution, we can detect vehicles at a greater distance and with more of a top-down view. This allows the camera to be mounted higher to avoid possible occlusions and increase accuracy.
See our detailed specification on how to set up a Single- / Multispace parking scene in terms of camera placement:

Speed estimation

The speed estimation feature allows you to extend your counting line to measure speed based on a configurable distance calibration. Speed estimation can be combined with any Traffic counting use case and allows you to get the count including the speed estimation for each object. In Data Analytics, you can then see the number of vehicles passing the CL in a given speed corridor or see the average speed on your aggregated counts. You also have the option to see the historical average speed of your speed estimation configuration over a period of time.


Outdoor ANPR - improvement as well as specification including detailed set-up guide

With this Release, we finally support ANPR use cases as well outdoor. The main challenge for ANPR setups consists of clearly readable license plates. This means a sharp and well-illuminated image without occlusions or blurry objects is required to obtain good results. In order to have the best set-up as well for outdoor installations, we have created a set-up guide with recommendations for your ANPR outdoor installations.
Improved License plate reading & support for all EU countries
Our ANPR feature got enhanced and supports all European license plates with the same accuracy. The accuracy was increased for any ANPR-relevant scene drastically. This improves reading the License plates more accurately, especially for countries besides Austria, Germany, Switzerland and Netherlands which have been already at the definded accuracy level before. License plates from Norway and Liechtenstein are supported now as well.

Traffic model improvements for better people & bicycle detection

Our Traffic model has been specifically improved for more stable detection of persons and bicycles. In addition, our Performance Lab has been extended to include the use case of measuring Origin-Destination of persons and bicycles at pedestrian crossings, in order to achieve even more reliable accuracy figures and more improvements in the detection of persons and bicycles in the future.

SWARM Control Center improvements

The SWARM Control Center has been improved in terms of usability and missing information according to your feedback. The following options are new:
  • All UUIDs are now shown in the SCC and enhanced with a dedicated copy button to copy them easier --> This will ease the integration to third-party systems as you can get the UUID directly out of the SCC.
  • Auto-Refresh option for device table in device configuration is now available. The whole table will be updated automatically every 10 seconds. So the connection and status and list of the devices will be up to date at any time without a need of refreshing the page.
  • In case you have changed a configuration in your scenario configuration and you accidentally close the tab you will be asked if you are sure to close it as you will lose the configuration change.
  • The filters which are applied at the table will be persisted on your browser. So any action you take the filter will stay. In order to see that there is a filter applied there is a note and an option to reset the filter on the top of the table.

Data Analytics - widget loading time improvements

Out of internal tests and feedback from some customers, we realized that the initial load of the data for widgets took really long. We have improved this by loading pre-aggregations in the background regularly so that the data call for the widgets will load quicker.