Skip to content

README.md

Latest commit

 

History

History
35 lines (26 loc) · 725 Bytes

README.md

File metadata and controls

35 lines (26 loc) · 725 Bytes

PixelLink

This code is based on PixlLink and PSENet, the performance is not satisfactory

Requirements

  • Python 3.6
  • PyTorch v0.4.1+
  • opencv-python 3.4

Training

CUDA_VISIBLE_DEVICES=0,1,2,3 python train_ic15.py --arch vgg16 --batch_size 24

Testing

CUDA_VISIBLE_DEVICES=0 python test_ic15.py --resume [path of model]

Eval script for ICDAR 2015

cd eval
sh eval_ic15.sh

Performance

Dataset Pretrained Precision (%) Recall (%) F-measure (%) FPS (1080Ti) Input
ICDAR2015 No 81.2 75 78 5 1280*768

TODO

  • Find the bug of the performance issue.
  • Accomplish the code with better config file and more datasets