Repository files navigation

note: some code is inherited fromMhLiao/DB

中文解读

2020-06-07: 添加灰度图训练，训练灰度图时需要在配置里移除dataset.args.transforms.Normalize

conda env create -f environment.ymlgit clone https://github.com/WenmuZhou/DBNet.pytorch.gitcd DBNet.pytorch/

or

conda create -n dbnet python=3.6conda activate dbnetconda install ipython pip# python dependenciespip install -r requirement.txt# install PyTorch with cuda-10.1# Note that you can change the cudatoolkit version to the version you want.conda install pytorch torchvision cudatoolkit=10.1 -c pytorch# clone repogit clone https://github.com/WenmuZhou/DBNet.pytorch.gitcd DBNet.pytorch/

• pytorch 1.4+
• torchvision 0.5+
• gcc 4.9+

TBD

Training data: prepare a texttrain.txt in the following format, use '\t' as a separator

./datasets/train/img/001.jpg./datasets/train/gt/001.txt

Validation data: prepare a texttest.txt in the following format, use '\t' as a separator

./datasets/test/img/001.jpg./datasets/test/gt/001.txt

• Store images in theimg folder
• Store groundtruth in thegt folder

The groundtruth can be.txt files, with the following format:

x1, y1, x2, y2, x3, y3, x4, y4, annotation

1. config thedataset['train']['dataset'['data_path']',dataset['validate']['dataset'['data_path']inconfig/icdar2015_resnet18_fpn_DBhead_polyLR.yaml

• . single gpu train

bash singlel_gpu_train.sh

• . Multi-gpu training

bash multi_gpu_train.sh

eval.py is used to test model on test dataset

1. configmodel_path ineval.sh
2. use following script to test

bash eval.sh

predict.py Can be used to inference on all images in a folder

1. configmodel_path,input_folder,output_folder inpredict.sh
2. use following script to predict

bash predict.sh

You can change themodel_path in thepredict.sh file to your model location.

tips: if result is not good, you can changethre inpredict.sh

The project is still under development.

only train on ICDAR2015 dataset

TBD

• mutil gpu training

1. https://arxiv.org/pdf/1911.08947.pdf
2. https://github.com/WenmuZhou/PANet.pytorch
3. https://github.com/MhLiao/DB

If this repository helps you，please star it. Thanks.