Performance degradation on custom model

[aziz-ullah-khan]

I am fine tuning the model on Japanese Language, while training the metrics are showing high accuracy and detection is good enough. When I load the custom saved model the performance is drastically decreased or even not detecting anything. Any reason?

I am following this path https://github.com/JaidedAI/EasyOCR/blob/master/custom_model.md but I am not sure what I need to change.

Thank you

[rkcosmos]

probably overfitting.

[aziz-ullah-khan]

probably overfitting.

@rkcosmos , Thanks for the comment but I am checking the model on the same data on which the model is trained. Means both training and testing data are same. Please suggest what might be the issue.

Thanks again!

[Arunavameister]

@aziz-ullah-khan Hello, have you found an answer to your problem?