number plate recognition

http://matthewearl.github.io/2016/05/06/cnn-anpr/

#Dependencies:

• [TensorFlow]
• OpenCV
• NumPy

Usage is as follows: the sun webset : (http://groups.csail.mit.edu/vision/SUN/)

1. download the sun dataset(36GB) (http://vision.princeton.edu/projects/2010/SUN/SUN397.tar.gz) or just download the SUN2012(7.5GB)

2. Extract ~3GB of background images from the [SUN database into bgs/. (bgs/ must not already exist.) ----./extractbgs.py SUN2012.tar.gz

3. Generate 1000 test set images in train/. (train/ must not already exist.) ---- ./generate.py 1000 create a 'fonts',which show 'fonts/UKNumberPlate.ttf' downloaded here.

4. Train the model. ----- ./train.py the process will write the weights to weights.npz.

5. Detect number plates in an image. -----./detect.py test/test1.jpg weights.npz out.jpg

your menu should like this: https://github.com/QueenJuliaZxx/number_plate-recognition/blob/master/Screenshot%20from%202017-11-22%2010:27:04.png