EasyPR-python by fh
slow, some mistake
- python 3.5
- opencv 3.1
- tensorflow 1.0
Use easypr-1.4's way to do locate and segment.
Use Lenet to do chars identify.
Use a cnn network to judge whether a car.
...
- plate locate ( need test )
- plate judge
- plate judge (cnn 98.79%)
- plate segment
- char identify ( cnn 99.63%)
unzip resources/train_data/chars.7z
python demo.py
(There is already a trained checkpoint in train/model/chars/models)
similar with identify
python demo.py