saicoco / tf-sotd

Based EAST implements "Self-organized Text Detection with Minimal Post-processing via Border Learning"

tf-sotd

tensorflow version "Self-organized Text Detection with Minimal Post-processing via Border Learning" <<<<<<< HEAD

feature

• support [x1, y1, x2, y2, x3, y3, x4, y4] format label
• integrate into east
• multitask training with east and sotd

Usage

1. Downloads Resnet_v1 from here
2. Organize your dataset
3. train: ./train.sh
4. test:

./test_single_img.sh

Some Results

I train the model on my diy-dataset, and I test it on crowd text, here is a sample result from "沉默的大多数"

References

• EAST
• sotd-keras-theano-version