Open Deep-21 opened 4 years ago
There are 2 character recognition models in the repositiory: 1) char_reg.hdf5 :- 90% 2) weights-improvement-03-0.93.hdf5 :- 93%
really great work ...plate detection is gud one , but character recognition from detected image.png not a gud , plz suggest me , how to improve the accuracy from character recognition part ..need for the train the model or add new pre trained weights for improving accuracy for recognition part ...
for retraining the character recognition model. you can use the train_detect.py scripts in the Training folder
really great work ...plate detection is gud one , but character recognition from detected image.png not a gud , plz suggest me , how to improve the accuracy from character recognition part ..need for the train the model or add new pre trained weights for improving accuracy for recognition part ...