search

Media-Smart / vedastr

A scene text recognition toolbox based on PyTorch
Apache License 2.0
534 stars 100 forks source link

Whether to try on the Chinese datasets? #3

Closed whuhangzhang closed 4 years ago

whuhangzhang commented 4 years ago

Thank you for your great work. Have you ever experimented with Chinese data sets? Can you provide some other configuration, such as CRNN algorithm, etc

ChaseMonsterAway commented 4 years ago

@whuhangzhang , hi We haven't experimented with chinese data sets. It's easy to implement CRNN by modifying tps-resnet-bilstm-attn configure file. We will implement other str algorithms in the future.

whuhangzhang commented 4 years ago

Thanks for your reply @ChaseMonsterAway . I will make some experiments on the Chinese dataset, so I hope you can give me some Suggestions. For example, whether the resnet-fc model can work for the text recognition classification of 6000+ words? Any Suggestions on preprocessing of text？thank you！

ChaseMonsterAway commented 4 years ago

It will take up lots of memory.