Is pretrained CRNN bias towards Chinese character? #64
Description
I used pretrained model and do a test on the t3.png available in the test_images folder. Result is:
Biometrics,Fusedand Secured
Todav.ZOLOzuses facialrecognitiomcombined withindustmy-leading
spoofdetectiontechnoloavtoensurehatvourcustomersidentitiesy
remain uncompromised.
Imtheuture,weladdother biometrics,behavioralanalvtics and auestions
oniv individualuserscan answer tofurther strengthenthecustomer
experience withoutadding timeor frustration.
I would like to ask if this is due to the fact that your train data is biased towards Chinese character? I noticed that there is no need to use spaces between Chinese characters in a sentence.