Please store pretrained model in Google Drive where we can easily access, not able to access via Baidu Netdisk

[Akshay-Ijantkar]

@ying09

[ying09]

@ying09

Ok, I will upload our pretrained models to Google Driver as soon as possible.

[Akshay-Ijantkar]

@ying09 Yes please, please try to do it as early as possible, great work on paper, by the way!

[ying09]

@ying09 Yes please, please try to do it as early as possible, great work on paper, by the way!

I have uploaded. https://drive.google.com/drive/folders/18Ll-3bAmi4CR2eGTuM-j6fkMrSAaBV4Z?usp=sharing

[Akshay-Ijantkar]

@ying09 Thank you! I also suggest please try to have installation steps to reproduce your results, especially requirements.txt, Thanks!

[Akshay-Ijantkar]

@ying09 I have saved weights in out_dir_r101 folder and then ran

python3 demo/icdar2013_detection.py  

I am getting this error:

    raise KeyError("Non-existent config key: {}".format(full_key))
KeyError: 'Non-existent config key: MODEL.TEXTFUSENET_MUTIL_PATH_FUSE_ON'

Can you please help me with it?

[Real-YeJ]

@Akshay-Ijantkar It seems that you didn't install the conda environment and build our TextFuseNet-detectron2 correctly.

[Akshay-Ijantkar]

@ying09 Steps which I followed for getting started with this:

!git clone https://github.com/ying09/TextFuseNet.git

! pip install torch==1.6.0+cu101 torchvision==0.7.0+cu101 -f https://download.pytorch.org/whl/torch_stable.html

!python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'

!pip install PyYAML==5.3.1

! python TextFuseNet/demo/icdar2015_detection.py  --input img540.jpg

Do I need to build it python setup.py build

then install it: python setup.py install

Please let me know! Thanks!

[Real-YeJ]

@Akshay-Ijantkar
first, you need to install the conda environment according to the installation of detectron2. I suggest that you configure it completely according to our requirements, because different versions may lead to unknown influence on performance. and then, you can build the TextFuseNet-detectron2 by running ' python setup.py build develop '.

[ying09]

@ying09 Steps which I followed for getting started with this:

!git clone https://github.com/ying09/TextFuseNet.git

! pip install torch==1.6.0+cu101 torchvision==0.7.0+cu101 -f https://download.pytorch.org/whl/torch_stable.html

!python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'

!pip install PyYAML==5.3.1

! python TextFuseNet/demo/icdar2015_detection.py  --input img540.jpg

Do I need to build it python setup.py build

then install it: python setup.py install

Please let me know! Thanks!

We will give the instructions of step-by-step installation as soon as possible.

[ying09]

@ying09 Steps which I followed for getting started with this:

!git clone https://github.com/ying09/TextFuseNet.git

! pip install torch==1.6.0+cu101 torchvision==0.7.0+cu101 -f https://download.pytorch.org/whl/torch_stable.html

!python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'

!pip install PyYAML==5.3.1

! python TextFuseNet/demo/icdar2015_detection.py  --input img540.jpg

Do I need to build it python setup.py build

then install it: python setup.py install

Please let me know! Thanks!

@Akshay-Ijantkar Hi, we have uploaded the instruction of step-by-step installation. Please refer to https://github.com/ying09/TextFuseNet/blob/master/step-by-step%20installation.txt

[Akshay-Ijantkar]

@ying09 Thank you! Will try that