The raspberry Pi ANPR Camera PiAnpr
Click here to stay updated with my project and see the kickstarter backers page ( If funded )http://jamiej7.wix.com/anpr-on-raspberry-pi
What I have done so far is build a working prototype of a raspberry pi powered ANPR numberplate recognition camera.
So far I am able to do the recognition on a remote pc with streams from the PiAnpr live from the built it router over wifi. My next step is to get the optical character recognition done on the raspberry pi itself and to send out the data via either ftp or http to a remote server
Your help will fund for components to build a fully working deployable camera.
With the funding it will enable me to build a new Camera learning from the issues with the prototype.
THE STAGES AND WHAT YOU FUNDS ARE BEING USED FOR
Stage 1, re design & rebuild the front and back panels to enable me Increase the illumination to 3 x 5w LUXEON 940nm led lamps plus drivers and heat syncs on the front . and to make the rear waterproof To also purchase some new infra red filters This will take about 2 days to build and complete.
Stage 2, I will be introducing a GSM network (connection over 4g) to the camera. tis will be within a remote enclosure with the remote power supply (see stage 5) This will take about 1-2 days with installing and testing
Stage 3, I will also be able to purchase and run Imps.tm LPR software coded to the Pi which has a licence of $500 dollars per Raspberry Pi. I will also be testing JavaAnpr to compare the two and writing scripts to handle the images from the camera and running them through the lpr engine and returning the results the images will be buffered in a folder and stored until passed to the engine this ensures all plates are captured and not missed This will take about 5 days with installing, testing and training the engine to recognise UK plates
Stage 4, add motorized lens ( controlling the lens via the GPIO outputs on the pi) and add the zoom and focus buttons to the rpi web cam interface This will take about 2-3 days with installing,
Stage 5, I will also Create a portable power supply inside a waterproof (ip66+) enclosue to give the camera true remote deployment. This will take about 2-3 days with building,
Stage 6, Add fixings and bracket to the camera and remote gsm/psu unit for pole mounting This will take about 2-3 days with building,
I plan to complete this project by mid august
If this project reaches over 100& of its goal, I want to buy a solar panel for long term deployment and if possible some new raspberry pi's to over-clock and test to get the best optical character recognition
Risks and challenges
Possible setbacks will include production of a cs mount for the pi noir and fitting of the lens.
Also getting such a high powered led chip from china may be a delay and creating an ip65 rated enclosure that will dissipate the heat from the chips
if this is the case a separate led chip may need to be mounted in a separate enclosure attached to the side of the main camera body
I have yet to see the results of optical character recognition BY IMPS.TM done directly on the pi apart from a short video done by their team with a standard USB webcam (the video is on Youtube). I have tested Javaanpr a little with good results and further tweeking to be done on training the engine. I also have no idea yet what the results will be from over-clocking the Raspberry pi in relation the the character recognition speeds.Learn about accountability on Kickstarter
- (30 days)