jameslimers / document_layout_analysis

Implementation code for document layout analysis (Hackathon 2020 in Suzhou)

Document_layout_analysis

Implementation code for document layout analysis

This work is implemented by "Machine fancy behavour Lab" Team

Presentation(Chinese):

link: https://pan.baidu.com/s/1qk9LP3zi6xX4V2sB-3PBiQ code: dumw

PPT(Chinese):

link: https://pan.baidu.com/s/1kgwZ5_-apCpG-W_EnuPW-g code: q1kv

contect us:

• [email protected]
• [email protected]
• [email protected]

Contents

1. Environment Setup
2. Demo
3. Training Models

Environment setup

• Clone the repository

git clone https://github.com/JaMesLiMers/Hakathon_layout_analysis.git && cd Hakathon_layout_analysis

• Setup python environment

conda create -n torch python=3.6
source activate torch
pip install -r requirements.txt

Demo

• Some of ouf model structure and performance

• Setup your environment
• Download the trained model and put into Hakathon_layout_analysis/save/
• trained model avaliable in:

https://drive.google.com/open?id=1YSNEL5xzaLlfLiU7t1sEnSGPa3SI9-3O

• Put input image into Hakathon_layout_analysis/test/Image('.jpg' file)
• Run test.py/ test_second.py

python ./test.py --resume model_epoch_409.pkl
python ./test_second.py --resume model_second_epoch_409.pkl

• Output will be generated in Hakathon_layout_analysis/test/Output

Training Models

• Setup your environment
• Download test data(Not avaliable now) put into Hakathon_layout_analysis/Data/
• Download pretrained models(alexnet_bn_.pth), avaliable in:

https://drive.google.com/open?id=1YSNEL5xzaLlfLiU7t1sEnSGPa3SI9-3O

• Run train code:

python ./train.py
python ./train_second.py

• Output will be generated in Hakathon_layout_analysis/save/

document_layout_analysis's Issues

如何获取训练数据集

dataset

请问您能提供训练的数据集吗？

how to get trained model?

how to get trained model? please put trained model to baidu's pan.
国内好像获得不了训练的模型。能否传到百度网盘上。