Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network

Requirements

• pytorch 1.1
• torchvision 0.3
• pyclipper
• opencv3

Data Preparation

prepare a text in the following format, use '\t' as a separator

/path/to/img.jpg path/to/label.txt
...

Train

1. config the train_data_path,val_data_pathin config.json
2. use following script to run

python3 train.py

Test

eval.py is used to test model on test dataset

1. config model_path, img_path, gt_path, save_path in eval.py
2. use following script to test

python3 eval.py

Predict

predict.py is used to inference on single image

1. config model_path, img_path, in predict.py
2. use following script to predict

python3 predict.py

The project is still under development.

Performance

ICDAR 2015

only train on ICDAR2015 dataset

Method	Precision (%)	Recall (%)	F-measure (%)	FPS
paper(resnet18+short_size:736)	x	x	80.4	26.1
my implementation(resnet50+short_size:736+pse扩张)	60.06	48.57	53.71	12.18 (P100)
my implementation(resnet50+short_size:736+聚类)	54.91	45.21	49.59	1.44 (P100)
my implementation(psenet+resnet50+short_size:736+pse扩张)	76.9	78.57	77.73	8.79 (P100)

examples

TBD

reference

1. https://arxiv.org/pdf/1908.05900.pdf
2. https://github.com/WenmuZhou/PSENet.pytorch