AnyTrans: Translate AnyText in the Image with Large Scale Models

🎉🎉🎉 This is a brief implementation of AnyTrans proposed by our paper ["AnyTrans: Translate AnyText in the Image with Large Scale Models"].(EMNLP2024 Findings)

Since our framework is quite simple, we only provide a general framework, and users can add LLM or erase tools according to their needs.

Datasets

all2zh dataset GoogleDrive

zh2all dataset GoogleDrive

Visualization