OpenCV_RecognizingCars

Using Machine Vision to recognize license plates, make and model on a car for COMP4102 project.

Demonstration

https://www.youtube.com/watch?v=pPQLGuZBYxk

LPR Instructions

Ensure you have the dependencies installed!

pip install opencv-python numpy matplotlib imutils 
git clone https://github.com/godrowr/OpenCV_RecognizingCars
cd OpenCV_RecognizingCars
gedit LPR-Haar-Contour.py

Change the path to the file you wish to preform license plate recognition.

• Contours work with all images in LicenseImages except 12.png.

path = "LicenseImages/1"
algo = ".png"

• Haar cascade "russian_plate_number.xml" works for only suzuki_car.jpeg

path = "LicenseImages/suzuki_car"
algo = ".jpeg"

• Haar cascade "cascade.xml" works only for 12.png.

path = "LicenseImages/12"
algo = ".png"

Haar video algorithm runs with the mp4 clip included.

Then uncomment out the line of the method you wish to run.

	haar_image_detect(imgGray,plate_cascade)
	#plate, frame  = haar_video_detect(russian_plate_number_cascade)
	#contour_image_detect(imgGray)

Then run the LPR code.

python LPR-Haar-Contour.py

OCR Instructions

python segmentation.py or seg.py “your image file name” to segment your image into single character.

''' An example would be passing the program 'example.png' or 'example2.png' incase other license plates don't segment correctly from the licensePlates folder images. The result images would write to the “Output” folder. '''

To run shape context algorithm for segmented characters: python shape_context.py to perform shape context recognition once you have segemented characters running the segmentation.py above.

''' It will reads all character images and display the prediction '''

To run shape context algorithm for testing with training data: python SHAPE.py

''' Pass in a index between 0-653 on line 353 and line 378 for the label to print the exact value and predicted value. '''

References:

Special thanks to all these people!

Louka Dlagnekov (2005). Video-based Car Surveillance: License Plate, Make, and Model Recognition. University of California. Available at: http://vision.ucsd.edu/belongie-grp/research/carRec/dlagnekov_thesis_2005.pdf

Russian Car Crash compilation of road accidents, YouTube. Available at:

https://www.youtube.com/watch?v=nHVY5TTMoyY

Naotoshi Seo - Tutorial: OpenCV haartraining (Rapid Object Detection with a Cascade of Boosted Classifiers Based on Haar-like Features). Accessed 04, 2021. Available at: http://note.sonots.com/SciSoftware/haartraining.html

Plates Portal (Ontario), Accessed 03, 2021. Available at:

http://plates.portal.free.fr/canada/ontario-cdn.html

License Plate Mania (Canada), Accessed 03, 2021. Available at:

http://licenseplatemania.com/landenframes/canada_fr.htm

LICENSE PLATES OF THE WORLD (Ontario) by Michael Kustermann. Accessed 04,2021. Available at:

http://www.worldlicenseplates.com/world/CN_ONTA.html

Olav’s Plates (Ontario), Accessed 03, 2021. Available at:

Http://www.olavsplates.com/canada.html

Dutch Numberplate Archives. Accessed 03, 2021. Available at:

http://dna.nl/

Tutorial-haartraining, handaga. Github. Accessed 03, 2021. Available at:

https://github.com/handaga/tutorial-haartraining

OpenCV-Haar-Adaboost, Andreluizfc. Github. Accessed 03, 2021. Available at:

https://github.com/Andreluizfc/OpenCV-Haar-AdaBoost

OpenCV_3_License_Plate_Recognition_Python, MicrocontrollersAndMore. Github. Accessed 03,2021. Available at:

https://github.com/MicrocontrollersAndMore/OpenCV_3_License_Plate_Recognition_Python/tree/master/LicPlateImages

Shape Context Matching For Efficient OCR

https://people.csail.mit.edu/spillai/data/papers/scocr-project-paper.pdf

Shape Context descriptor and fast characters recognition

https://medium.com/machine-learning-world/shape-context-descriptor-and-fast-characters-recognition-c031eac726f9

Shape context Matching descriptor

https://en.wikipedia.org/wiki/Shape_context

Shape Context: A New Descriptor for object matching and object recognition

https://www.researchgate.net/publication/2353535_Shape_Context_A_new_descriptor_for_shape_matching_and_object_recognition/link/0912f50b3498fd04df000000/download