fujingling / DRRG

Deep relational reasoning graph network for arbitrary shape text detection; Accepted by CVPR 2020 (Oral). http://arxiv.org/abs/2003.07493

This is an implementation of our CVPR2020 paper. The complete code will be provided after returning to school. Please wait patiently!

1.Prerequisites

python 3.7;
PyTorch 1.2.0;
Numpy >=1.16;
CUDA 10.1;
GCC >=9.0;
NVIDIA GPU(with 10G or larger GPU memory for inference);

2.Description

• Generally, this code has following features:
  1.Just include complete inference code
  2.Support TD500 and CTW1500 datasets

3.Parameter setting

• CTW1500: follow the model/Ctw1500/ctw1500_test.txt
• TD500: follow the model/TD500/TD500_test.txt

4.Pretrained Models

• CTW1500 pretrained model: CTW1500
• TD500 pretrained model: TD500

5.Running tests

• Preparation

1. git clone https://github.com/anoycode22/DRRG.git
2. put your test images in "data/TD500/Test" or data/ctw1500/test/text_image
3. put the pretrained model into "model/TD500/" or "model/Ctw1500"
4. cd ./csrc and make
5. cd ./nmslib/lanms and make

• CTW1500

1. set the parameter in config according to model/Ctw1500/ctw1500_test.txt
2. python eval_TextGraph.py --exp_name Ctw1500 --test_size \(512, 1024\)

• TD500

1. set the parameter in config according to model/TD500/TD500_test.txt
2. python eval_TextGraph.py --exp_name TD500 --test_size \(512, 640\)

6.Qualitative results()

References

(S. Zhang, X. Zhu, H.-J. Bo, C. Liu, C. Yang, H. Wang, and X.-C. Yin, "Deep relational reasoning graph network for arbitrary shape text detection", CVPR 2020).