What’s new in Watcher ver. 22.02

  1. In some cases, the operator should validate the detected number plate or identify some vehicle’s features. Since this release HD quality screenshots for ANPR events are available in the Web UI. You can also receive a screenshot via API.

  2. By default, the recognition system searches for license plates and faces over all the camera fields of view. In this release, we have added the ability to configure the detection area. Thus, you can select the specific polygonal area(s) to detect license plates and faces. This setting may help you to reduce false detections.
    As an example, you can have an automatic opening of the barrier by recognized license plates. In this case, you can select as the recognition zone only the part of the road that is located directly in front of the barrier. Thus, the system will not be triggered on other vehicles passing by.
    22.02

  3. We continue to improve video analytics. For example, we have reduced false detections: there is a more stable behaviour of the video analytics module regarding objects similar to a human face, but which are not.

What’s new in Flussonic ver. 22.02

  1. In one of the previous releases, adaptive bitrate support for WebRTC (ultra-low latency data transfer protocol) was added. We continue to improve WebRTC: starting from version 22.02, Flussonic retransmit lost UDP packets to a browser after getting the event that a packet was not received. This allows you to have fewer artifacts in the transmitted video if the internet connection is unstable. This works for UDP packets during WebRTC playback.
    Why WebRTC over UDP is useful if WebRTC over TCP already implies a built-in packet retransmitting mechanism? TCP implements loss compensation, but with high latency. In this release, we can say that we have made our own version of WebRTC TCP, but working over UDP (that is, faster). New frames continue to be sent to the client (player/viewer/computer) without confirming whether he has received the old ones. Packets are retransmitted only when the browser reports a loss. This is different from TCP, where confirmation of receipt of each packet is awaited.
    Ultimately, we managed to combine minimal latency and artifact-free video transmission even with an unstable internet connection. WebRTC over UDP mode should give a latency less for a time equal to RTT. This applies to both connection establishment and latency. (RTT is stable around 30-100ms).

  2. We added support for the AV1 codec. AV1 is an open-source (free of charge) codec, which is 2 times more efficient than h264 (according to this data). It compresses better and allows you to get better video quality at the same bit rate (and conversely, to use less bandwidth while maintaining the same quality).
    In particular, it became possible to publish and play videos encoded with AV1 via WebRTC. WebRTC imposes strict requirements for latency, so the computer has little time to compensate for lost data, to retransmit packets. In these conditions, effective use of the Internet bandwidth becomes important: the better the compression (and with the AVC1 codec it is better), the less data is transmitted over the Internet and the less loss of this data. Thus, with the same Internet, the video quality when publishing and playing via WebRTC will be better with the AV1 codec.