Modern cities are facing challenges like making cities smart and managing traffic. Cameras have proven to be efficient sensors for smart city applications. However, the need for use of different applications and hardware in various scenarios made the smart city dream a costly “integration hell”. Then DataFromSky came with FLOW, a smart traffic analysis framework which lets you turn a camera to any sensor you need. FLOW is able to process video with high accuracy and utilize trajectory driven design to generate precise traffic data. FLOW is a visionary AI based solution with many applications. It can improve traffic in the city, manage parking, improve safety with certified collision detection or gather data for retail. FLOW can recognize vehicle types, pedestrians and bicycles and measure values like speed, acceleration or timings with high precision and it can do this for 100s of cameras at once. But the best part of FLOW is how easy it is to use using the visual programming dashboard.

In this webinar we will introduce you to FLOW and the products that use it.

The webinar also has a live demonstration to show you how easily the measurement of different types of traffic data can be set up inside the intuitive visual programming interface.

What will you learn in this webinar?

1. Learn how to turn your cameras into multipurpose sensors for smart cities with the FLOW framework

2. Learn about modern city camera based solutions using DataFromSky footage analysis method and the traffic analysis framework FLOW

3. Learn about the different uses of FLOW (traffic, parking, retail, safety) and the FLOW family products types (Camera, Embedded (local multi-camera processing unit), Enterprise (server processing))

4. Learn how to set up traffic data measuring in FLOW in 20 minutes (example of city junction) - try FLOW as you watch.

You can find the webinar timeline in the video description on VIMEO.

Questions and Answers

30:30 Q1: What is the accuracy of the vehicle detection?

A: Accuracy > 98% . Depends on the input of the flow Engine (the camera resolution), camera position and framerate needed. Necessary frame rates vary - 15FPS in the city. Highway needs more, some 20-25 FPS or more due to higher speed of vehicles. 32x32 pixel detection resolution is needed for an object to be recognised. Accuracy can be upgraded to 100%.

31:55 Q2: How many detection categories are available?

A: Standard mode has 7 classes: car, light vehicle, heavy vehicle, bus, motorcycle, bicycle and pedestrian. Can be extended to up to 16 classes based on the needs of a project. These classes can be specifically counted in the analysis - for example you can detect number of heavy trucks passing through a given area.

32:30 Q3: What Hardware platform do you use?

A: NVIDIA Jetson is used in Embedded. NVIDIA GPUS in Enterprise. Embedded Nano uses Jetson Nano and Embedded Micro and Macro use Jetson Xavier. FLOW is also available on Azena app store. Various FLOW based apps can be found on the store including FLOW Traffic, FLOW Retail, FLOW Security and FLOW Parking. The camera hardware depends on what camera you choose. We can help you decide what camera to buy and where to position it for optimal results.

33:15 Q4: Can you measure distances between people?

A: Yes we can. We can measure distances between objects with up to 30 cm precision. This can be even more accurate based on the video quality. Our data was used in many research projects where precision is very important. If you get in touch with us we can offer a tailored solution with very precise data.

34:08 Q5: Can I use FLOW for traffic studies from video recordings?

A: No. FLOW is for live processing only but we have the TrafficSurvey product which is designed for this purpose. We are able to analyse large areas and track the whole trajectories of objects within these areas. For example we did a traffic survey for a 1km long area covering multiple junctions. You can hear more about it in this Webinar.

The project is presented from 45:30.

34:47 Q6: Is there a way to try/test FLOW?

A: Yes - here. The FLOW demo kit for can be downloaded for free. Inside it, you will find a traffic simulation you can try to analyse in real-time. It is currently available for Windows only - if you would like to learn more please Contact us!. Manuals and guides on FLOW found on our intercom.

Conclusion

In this webinar you have learned about the issue of smart cities and how it can be solved with FLOW. You have learned about the different reasons that make FLOW special such as the wide variety of applications, the scale on which it can operate, the ease of set up with existing camera infrastructure and the intuitive easy to use visual interface. If you would like you can read about more about core concepts of FLOW or the architecture of FLOW on our Help Center.

We showed you how to do a simple vehicle count, set up wrong way driving detection, U-turn detection, stationary vehicle detection with using gates. You have also learned how to make Widgets like counts or graphs for a customizable dashboard or how to set up Sinks to send the data from FLOW outside. We hope you liked it!

• Do you have more questions? Contact us!

• Follow us on Linkedin.

• Learn more about FLOW.

• Visit our homepage to view products and news.

Thank you for your interest. Have a great day and let the traffic flow - with FLOW!

Did this answer your question?