https://github.com/chinakook/CTPN.mxnet

Connectionist Text Proposal Network in MXNet
https://github.com/chinakook/CTPN.mxnet

ctpn ocr

Last synced: about 2 months ago
JSON representation

Connectionist Text Proposal Network in MXNet

• Host: GitHub
• URL: https://github.com/chinakook/CTPN.mxnet
• Owner: chinakook
• License: other
• Created: 2018-05-22T08:04:23.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-10-14T12:17:47.000Z (about 6 years ago)
• Last Synced: 2024-08-01T22:41:27.278Z (5 months ago)
• Topics: ctpn, ocr
• Language: Python
• Size: 2.32 MB
• Stars: 70
• Watchers: 8
• Forks: 19
• Open Issues: 3
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• Awesome-MXNet - **CTPN.mxnet**

README

        
# Connectionist Text Proposal Network in MXNet

## Introduction

CTPN is a nice scene text detection method.

[**中文文档**](./README_CN.md)

## Training

1. Execute script init.sh(init.bat on Windows) to initialize project.

2. Download pretrained model from [here](http://data.mxnet.io/models/imagenet/vgg/vgg16-0000.params) into ```model``` folder.

3. Download dataset from [google drive](https://drive.google.com/open?id=0B_WmJoEtfGhDRl82b1dJTjB2ZGc) or [baidu yun](https://pan.baidu.com/s/1kUNTl1l). This dataset is already prepared by **@eragonruan** to fit CTPN.

4. Unzip the dataset downloaded to ```'VOCdevkit'``` folder, and set both ```default.root_path``` and ```default.dataset_path``` in ```rcnn/config.py``` to ```'/VOCdevkit/VOC2007'```. You can also change other hyperparams in ```rcnn/config.py```.

5. Run ```python train_ctpn.py``` to train. Run ```python train_ctpn.py --gpus '0' --rpn_lr 0.01 --no_flip 0``` to train model on gpu 0 with learning rate 0.01 and with flip data augmentation.

## Testing

Use ```python demo_ctpn.py --image "" --prefix model/rpn1 --epoch 8``` to test.

## Our results

`NOTICE:` all the photos used below are collected from the internet. If it affects you, please contact me to delete them.















## Requirements: Hardware

Any NVIDIA GPUs with at least **2GB** memory should be OK.

## References

1. https://github.com/tianzhi0549/CTPN

2. https://github.com/eragonruan/text-detection-ctpn

## TODO

- [ ] Custom dataset preparation tutorial

- [x] Windows support

- [ ] Deploying network and c++ inference support