It is a bit similar to the node-red-contrib-openalpr-cloud that I developed a couple of years ago. But the openalpr github repository seems to be abandoned for already quite some time.
Some good to knows:
You can very easily get started: sign up at their homepage, and paste your token into my config screen. That is all ...
This new node uses deep learning behind the scenes. That you can 'learn' a piece of software to recognize something, really keeps fascinating me...
They offer a free license up to 2500 images per month, which is quite well for hobbyist IOT users.
The service can also be installed locally (via their SDK) on a number of platforms, like e.g. Raspberry.
They offer a Docker container, so you should be able to run it locally without having installation headache...
I have added a demo on my readme page, which should get you up and running very easily:
There is something I'm not really sure about.
You can now filter specific regions, by entering an array of region codes:
But not sure how it works.
In the documentation I see this:
I did a test last night with a german plate ("de"). Then I entered an array for only belgian and french plates ([ "be", "fr" ]), but it still recognized the german plate. Thought it would only report belgian and french plates?
If anybody knows how I should interpret the documentation, would be nice to let me know!
BTW beside license plate recognition, they also offer an API to query your statistics. E.g. how many recognitions you still have left this month in your free account, and so on ...
Can imagine that it would be useful to get this information, e.g. to show an alert on your mobile file when you are running out of recognitions.
But should I create a separate node in my repository, or add it to my existing node (e.g. when some topic arrives) ... Any suggestions?
I want to congratulate you on a job well done, and for full usage for the nodes.
I also want to thank you for helping out on this project, and I am sure that this will be the project I will refer to when I am stuck on a coding.
And I promise to give to the community my full potential.
Once again thank you for your cooperation and have a nice day...
Do you think it would be worthwhile to add a message to the node-red-contrib-openalpr-cloud readme, advising users to install this node instead?
I haven't as yet tried the new node, but was very impressed how effective platerecognizer.com was in recognising poor quality images.
Yes I am also impressed about it. And yes I could also add a link to the readme page. This also reminds me that I need to add the keyword "alpr" to my package.json file, to guide users towards my new node. Now off to work...
I added the two remaining functions from the docs page (send timestamp and cameraID), I forked your repository and made changes, also made some changes to the pieces of information, when you are available please check them out, would love to push them also to npm
Yes that is true ... Indeed I 'think' both fields have no impact on the recognition in the service.
But now at least the config screen looks a bit impressive
I think what Dave was saying was that you could have the timestamp feed without having a selectable option in config.
It would be there all the time, but if you didn't want to use it, then disregard it.
Having as a selectable option just adds unnecessary clutter to the node config, and really serves no practical purpose.
But hey, it's yours & Bart's node, so it's your preference that matters.
Good point, so @lizzardguki has implemented a PR to solve this. Now we always send the timestamp in version 1.0.2:
The checkbox removal has not really reduced the size of the config screen very much, so this change hasn't enabled me to buy a smaller smartphone and still fit the config screen on it
This new 1.0.2. version also has improved error handling. I had a party last night at home, so lots of cars from my international friends on my driveway. As a result my camera has sent this screenshot image to the plate recognition service:
But that seemed to be a bit too much for the recognition service. It responded with an internal server error, which caused our node to crash. Now we handle the unexpected result better:
Do you think we need a small changes to help all the new newbies on node-red.
My suggestion is for msg.payload to contain only the plate string.
If there are more plates it should list in msg.payload[ 0, 1,2 , etc]
What do you guys think about this?
