i want to create a license plate detection for iranian license plate that use european standard. for this work i want to use haar feature for localization and tesseract library for ocr and detect character. i collect 800 positive image that include license plate. and use 8 image that license plate in this picture removed.
after a hard work i extract 800 license plate with my hand nad train "trainCascade" in opencv but after testing, result is very bad that or detect any license plate or detect most license plate.
my sample is like follow:
for negative:
and
positive:
i have Question: 1-how to train for get best result(with traincascade in opencv)? 2-what ocr library is there for persian that recognize character in iranina license plate? 3-is there any project that may help me?
thaks.