[Detector Support]: Google Coral EdgeTPU + European Licence Plate dedection + Store?

[nousername1]

Describe the problem you are having

Hi
I am wondering if it work that I can record Licence Plate Information from an hight frame Rate USB Camera?
I bought an USB Cam with 620fps (depend on the resolution I need to use) so it should be enough to get the Licence Plate even at hight speed. Currently I could not test them at the place where the should be later.
The Decoder Software should only run when there is a Licence Plate in the Frame to reduce the CPU load.
I also would get an Google Coral EdgeTPU for my Server.
Thanks

Version

none

Frigate config file

none

docker-compose file or Docker CLI command

none

Relevant Frigate log output

none

Install method

Proxmox via Docker

Object Detector

Coral

Screenshots of the Frigate UI's System metrics pages

none

Any other information that may be helpful

No response

[hawkeye217]

Frigate already supports license plate detection with models that natively detect license_plate objects. Frigate 0.16 will include native license plate recognition (recognizing the characters on the plate).

With that said, a high frame rate camera is not necessary. Resolution is more important, as you want to capture the clearest image possible. You can configure Frigate's detect fps value appropriately. A max of 10fps is the recommendation for objects that move quickly across your frame.

In development and testing of native LPR cameras for Frigate 0.16 where plates move across the frame in ~1 second, 10fps has been more than sufficient.

[nousername1]

Thanks for replay. By law I cant record any Picture but I can record the Licence Plate Information.
Here you can see all the different Licence Plate Types in Europe:
https://de.wikipedia.org/wiki/Datei:Vehicle_registration_plates_in_Europe.png
Like with our Licence Plates: https://de.wikipedia.org/wiki/Kfz-Kennzeichen_(%C3%96sterreich) you can read out a lot of informations about the usage of the Car and where it is register.
All that should be stored with an 1sec. correct Time into an Database.

[hawkeye217]

Yes, you can configure Frigate to only detect and not record.

In Frigate 0.16, the characters of recognized license plates will be stored in Frigate's database and can be searched quickly.

[nousername1]

Is there a Video, Blog Post or what ever to dive deeper into?
Currently I need to buy some Stuff for the Cam. Need a location where I can "Wife approved" put the Cam on a tripoad and so on.
Later I want to test them when I convert my ESXi to an Proxmox System.

[hawkeye217]

You can look at the dev documentation, but Frigate 0.16 is still in development so you won't likely find any tutorials or blogs yet.

https://deploy-preview-16390--frigate-docs.netlify.app/configuration/license_plate_recognition/

There is a small segment of a user video discussing license plates here:

https://youtu.be/MYTr5W1GYYw?si=gchsN6GtOnsVkfrP&t=772

[NickM-27]

The Decoder Software should only run when there is a Licence Plate in the

How would you know (detect) that a license plate is present without decoding?

Frigate is an nvr that constantly monitors so it will always decode. Also, high fps isn't that important for license plate reading, I don't think a USB camera will do well https://www.reddit.com/r/homesecurity/comments/1ai0qlp/best_camera_for_seeing_license_plates/?rdt=41885

[nousername1]

By running truth a Google Coral EdgeTPU who is very very low power and when the dedect an Licence Plate a separate detection Software should start search the Image (or if the TPU can send the coordinates of the Licence plate it would be more power saving) for the Licence Plate.
My USB Cam is a hight frame Rate full Frame Camera in Monochrome when I test them out 2h ago that thing is a beast (for the Price)!
The Framerate is top! Well the resolution is not "that" important the must cover the Street with.

[hawkeye217]

Frigate (with the appropriate model) will detect the license plate, no need for separate software.