PPOCR-V3 RECOGNITION AUGMENTATION

[bely66]

In the paper of ppocrv3 they mentioned textconaug This augmentation concat 2 different images in a batch to form new images this augmentation is activated when i train

I have multiple questions

1. Does this mean two words will be concatenated from the dataset?

2. If yes, will the two words create a new word(concat with no space) or create a new sentence(concat with space)?

I'm training on word level so if concating adds space that might cause training issues