search

GXYM / TextBPN-Plus-Plus

Arbitrary Shape Text Detection via Boundary Transformer；The paper at: https://arxiv.org/abs/2205.05320, which has been accepted by IEEE Transactions on Multimedia (T-MM 2023).
172 stars 37 forks source link

How to train myself datasets? #26

Open AXNing opened 10 months ago

AXNing commented 10 months ago

I want to train and eval myself datasets,can you tell me what should i do please? @GXYM

GXYM commented 9 months ago

I want to train and eval myself datasets,can you tell me what should i do please? @GXYM

https://github.com/GXYM/TextBPN-Plus-Plus/tree/main/dataset and https://github.com/GXYM/TextBPN-Plus-Plus/tree/main/data. Here are various loading scripts for datasets in different formats. You can convert your data into any of these formats, or you can write a separate script to load your data by following these scripts yourself.

AXNing commented 9 months ago

I want to train and eval myself datasets,can you tell me what should i do please? @GXYM

https://github.com/GXYM/TextBPN-Plus-Plus/tree/main/dataset and https://github.com/GXYM/TextBPN-Plus-Plus/tree/main/data. Here are various loading scripts for datasets in different formats. You can convert your data into any of these formats, or you can write a separate script to load your data by following these scripts yourself.

Thanks,i can train my dataset.But i find i can not remember the train loss,i do not find train logs.Please tell me where i can remember the train logs.Thank you very much.

refer to https://github.com/GXYM/TextBPN-Plus-Plus/issues/25