README.md

---
license: apache-2.0
tags:
- scene text erase
- poster text erase
---

# Self-supervised Text Erasing Model (STE)
Paper: [https://arxiv.org/abs/2204.12743](https://arxiv.org/abs/2204.12743)<br/>
Project Page: [https://github.com/alimama-creative/Self-supervised-Text-Erasing](https://github.com/alimama-creative/Self-supervised-Text-Erasing)<br/>

## Description
The checkpoints are trained from the posterErase dataset. There are two versions with different training mechanism.

Self-supervised Text Trasing (ste_best_net_G.pth): To use it, please download from this page, and put it under './checkpoints/erasenet/ste/best_net_G.pth'

Finetuning after STE (ft_best_net_G.pth): To use it, please download from this page, and put it under './checkpoints/erasenet/ste/best_net_G.pth'

## Usage
First, download the github project and install the python package.
```bash
git clone https://github.com/alimama-creative/Self-supervised-Text-Erasing.git
pip install -r requirements.txt
```

Then, follow the command line provied in the github to run the inference code.

```bash
python test.py --dataset_mode items --dataroot ./examples/poster --model erasenet --name ft --which_epoch best  # inferece with the ste model on poster 

python test.py --dataset_mode items --dataroot ./examples/poster --model erasenet --name ste --which_epoch best # inferece with the finetuned model model on poster 

```