The FLOW 1.13 comes with several new features as well as improved video analytic performance. The most important update is the integration of new video detection and tracking engine that provides 23.8 % higher detection performance than the previous version and is 1.4x faster. The most exciting new feature is the traffic lights state detection which is now an integral part of the FLOW video analytical engine. Based on your requests, we have also added SSL certificate support on the BLOCK level.

Moreover, we have improved our UDP communication options by adding an object list sink as well as a better zone sink that can communicate object IDs and a UDP server that allows you to fragment the data as needed. Another useful addition to your arsenal is the easy License plate import for whitelisting and blacklisting specific license plates with ease and some other minor changes like the option to pause the analytics or see behind the scenes of the spatial filters with discretization policy. Now let's get to the nitty gritty details of these changes!

Traffic lights state detection

This is a highly anticipated feature that can be achieved without intrusion into the traffic controller data thanks to video analytics. Thanks to the new built-in traffic light status detector, you easily define the location of interest in the camera with the traffic lights view and then you use the recognized status for conditional evaluation of the traffic situation. A great application example is red-light runner detection which may be used for vehicles as well as pedestrians. More information in the article here!

With FLOW’s trajectory-based tracking you will be able to get all the data on the vehicle that you have gathered before it has run the red light including license plate and vehicle color which are helpful in identifying the specific car that has committed the violation. There are also other ways this function can be used.

By detecting the cars entering the intersection when yellow light or red light is on you can also help to prevent accidents by prolonging the red light signal slightly for the conflicting directions to prevent collisions. The same can be done for pedestrian crossings on the road.

FLOW has a built-in ALPR engine for reading and processing LPs since 1.11. But the manual definition of white-listed or black-listed LPs in License plate filtering operator starts to be time-consuming if you need to do it for more vehicles. This was addressed by a new function that allows you to import a list of LPs from a TXT file where LPs are separated by lines. Take a look at a short demonstration of using this feature when defining a parking monitoring task with different rules for known and unknown vehicles (check our how-to guide).

This new feature requested by our clients allows you to upload your own SSL certificate into the device via FLOW Insights. Thus you can use your own trusted certification authority to protect your data. This function is available in the Block settings under the section named “SSL webserver certificate for REST API”. If you upload an expired or self-signed certificate, Insights will notify you whenever you try to connect to the device.

One of the typical uses of the FLOW devices are real-time traffic monitoring tasks for dynamic control where the UDP sinks are preferred for communication with traffic controllers due to its low latencies and lightweight payloads. So far, FLOW supports only two options i.e. Zone sinks and Classification count. But with FLOW 1.13, you can start receiving the list of objects thanks to the new Object list sink. Moreover, we have also improved the existing Zone sinks which newly contain the list of trajectory identifiers, thus the receiving party knows which objects are presented in the zone. Besides that, we also added the configuration page for the UDP server where you can define the port and max. size of the datagrams (when enabled, the server fragments the data as requested). See more information here.

FLOW 1.13 comes with a lot of new visualization options. In the streaming interface, you can render the outputs of the video tracking engine directly to the mjpeg stream. The next great thing is that you can visualize this output on the dashboard using an inspection stream widget (be aware that mjpeg stream is not synchronized with other data on your dashboard, but the latency should be low). The streaming can be enabled in the Interfaces - Video streams section.

You can also freeze the image in the analytic definition to check the situation or you have now option to visualize the discretization grid to understand better how the framework evaluates spatial filters. The pause button can be found at the bottom of the button cluster inside the analytics view, right next to the button for visualizing the discretization grid. FLOW is now also available in Brazilian Portuguese and additional language versions are on the way. You can change the language from the main menu by clicking on the globe in the top right corner.

If you have a FLOW device with built-in ALPR, you may decide for each analytic individually if the ALPR engine will be used or not in the analytics settings. By enabling the ALPR engine on the analytics where LP recognition is required, you can save the computation power for other tasks and get higher processing FPS. Another new feature is that you can now define the sampling frequency for the CSV file export allowing you to export traffic data on a regular basis.

See the complete list of changes and new features here. The FLOW 1.13 version is available on all FLOW product lines: Cameras, Embedded, XRoads, Enterprise and Drone.

• Download FLOW demokit

• Do you have questions or feedback? Contact us!

• Follow us on Linkedin.

• Learn more about FLOW.

• Visit our homepage to view products and news.

• Check out our webinars.

• Read from our BLOG.

Thank you for your interest! Have a great day!

Let the traffic flow - with FLOW!