|Openalpr - An Open Source Licence Plate Reader|
|Written by MIke James|
|Friday, 18 December 2015|
Openalpr isn't a new project, but its latest release seems to have attracted some attention because of what it could be used for.
Openalpr is an open source licence plate reader that brings the sort of surveillance possibilities to almost any situation that once would have been restricted to law enforcement. Put simply, dedicated licence plate readers are expensive whereas software and standard IP cameras are cheap.
The first thing to say is that that as well as the open source software there is also a commercial cloud-hosted version which costs $50 per month per camera. The commercial version has a few extras, including a web app and plate groups. However, the open source version is enough to let you test out a system to see if it has any value for you and you can modify it to become part of your own app - as long as it too is open source.
It is a C/C++ application that can be run from the command line. There are also bindings for C#, Python, Node.js and Java. Using it is as easy as giving the path to the image file or using an in memory stream. It returns a list of licence numbers it found. It can accommodate the different styles of plate found in different parts of the world.
There is also an OpenALPR agent tat can run as a Linux daemon. In this mode it can monitor one or more MJPEG video streams and return JSON packaged data containing the licence number it found.
It isn't a neural network approach but an application of computer vision techniques. It makes use of OpenCV and the Tesseract OCR libraries. There is a nice explanation of how it works on its GitHub page. It would make a really good example if you are organizing a course on computer vision. It also provides plenty of scope for tinkering and the project welcomes any help you might want to put back into it.
There are binaries for Linux and Windows and a step-by-step account of how to compile it for Android and OSX.
What could you use it for?
The commercial web site is full of examples. Discover who is parking in your lot and do something to stop people who shouldn't be. Mange parking times for customers or employees. Keep track of who comes and goes to a facility and so on. All reasonable but some might find the example given in the video slightly disturbing:
The idea that existing IP cameras can be switched to licence plate readers does seems to pose problems for people worried about privacy. A camera trained on a street could give anyone all of the comings and goings in detail. It could be used to monitor access to wealthy city districts and for direct private security. However, without access to the criminal records of people associated with the plates it is difficult to see how effective this is going to be.
or email your comment to: firstname.lastname@example.org
|Last Updated ( Friday, 18 December 2015 )|