Synthetic-to-Real-Unsupervised-Domain-Adaptation-for-Scene-Text-Detection-in-the-Wild

Introduction

This is a project concerning Synthetic-to-Real Unsupervised Domain Adaptation for Scene Text Detection in the Wild

we propose a synthetic-to-real domain adaptation method for scene text detection, which transfers knowledge from synthetic data (source domain) to real data (target domain)

From more details,please refer to our paper

Environments

python 3
torch = 1.1.0
torchvision
Pillow
numpy

Train

The pre-trained model can be obtained from Baidu Drive password: n74a

Before taining the model, you need to configure related parameters:

resume
target_pseudo_negative
target_pseudo_positive
target_image
path_save

Our experiments are based on one machine with 2 2080ti(16g memory).

python trainSyndataToICDAR15.py

Eval

for example (batchsize=2)

CUDA_VISIBLE_DEVICES=0 python -u eval.py

License

For academic use, this project is licensed under the Apache License - see the LICENSE file for details. For commercial use, please contact the authors.

Citations

Please consider citing our paper in your publications if the project helps your research.

Eamil: weijia_wu@yeah.net

weijiawu/SyntoReal_STD