license-plate-recognition

the code for our CSC420 project

English

This directory contains the cleaned up version of CHARS74Kdataset we used to train our CNN for classification.

alphanumeric

This directory contains the dataset I put together for the detection CNN

detect_segment.py

the code for license plate detection and segmentation.

helper.py

Modified code from pytorch tutorial for display of CNN prediction output

model.py

The CNN models we used

train.py

The training and evaluation of both CNN.

train1.pt train2.pt train2v2.pt

The saved weights of the CNN. train1 is for detection CNN. train2 and train2v2 are for classification CNN train2 is trained on unmodified dataset. train2v2 is trained on augmented (using randon transformations, perspectives, etc)dataset. train2 generally works better.

example.png

An example image we used.

to run the example

run: python detect_segment.py 'example.png'