search

NinV / Korean-License-Plate-Recognition

Korean car license plate recognition using LPRNet
23 stars 3 forks source link

How can I get a graph of accuracy? #2

Open wkdcn97 opened 4 years ago

wkdcn97 commented 4 years ago

Hi, how can I get a graph of accuracy?

NinV commented 4 years ago

The "evaluate" method of class Evaluator returns CTC-loss, CER and WER. where WER is Word Error Rate. The whole text of a license plate is considered as a word so to you can get the accuracy by using (1 - WER). To get a graph of accuracy, you have to modify the "train" function in the train.py where the "evaluate" method is called.

mukul1em commented 4 years ago

Hi what was the accuracy of the model?

also do we have to label the image??