For Indian Number Plate I am getting blank space as output

[tallamtharunsai]

I have used the plain data like black and white image for training the model.
The training loss is also very high. 15

and i have used the pattern for the eval, that is : '^[A-Z]{2} [0-9]{2} [A-Z]{2} [0-9]{4}$' for this.
I am getting the blank spaces as output

Then i have aging checked if the whole thing is working or not. So, i have tried for pretrained Chinese data and the outputs of it are like this.

 

 

All the packages are uptodate

[IntelSupport]

Hi Tallamtharunsai,

Thank you for reaching out. Could you provide me more details on what program you are running? Which sample or demo of OpenVINO Toolkit you are running? Also, please provide me the environment of your machine and the version of OpenVINO Toolkit.

Regards,

Aznie

[tallamtharunsai · Accepted solution]

Sorry Sir,

My bad. I have done it with different batch size but kept the learning_rate and grad_noise_scale same. So, the model is not converged. I appropriate your reply for me. 

Thanks for the your time.

Regards

Tallam Tharun Sai.

[IntelSupport]

Hi TallamTharunSai,

Thank you for your question. If you need any additional information from Intel, please submit a new question as this thread is no longer being monitored. 

Regards,

Aznie