hyeongseokson1 / kpac Goto Github PK

Hi

Thank you for providing the code for your paper.

I am trying to replicate the results of the paper and would like to know which tensorflow version you use.
I have tried testing with tensorflow 1.10 and run into the following error with python main_eval_3level.py

[TL] UpSampling2dLayer sha_attention/0: is_scale:False size:[140.0, 210.0] method:1 align_corners:False
TypeError: Expected int32, got 140.0 of type 'float' instead

Thanks

Thank you for your excellent work! When we used the pre training model provided by you to evaluate, we found that the effect was very poor. The index of PSNR was only 5. The main problem encountered is that there are many parallel color artifacts in the evaluated image.
We wondered whether it was the problem of the installation environment. The environment we have configured is as follows.
Or can you provide us with pictures for your evaluation？Thank you!
The environment of python and tf is based on the version you provide.

_tflow_select 2.1.0
absl-py 0.15.0
astor 0.8.0
backports 1.1
backports.functools-lru-cache 1.6.4
backports.weakref 1.0.post1
blas 1.0
c-ares 1.18.1
ca-certificates 2022.07.19
certifi 2020.6.20
cloudpickle 1.3.0
cudatoolkit 9.0
cudnn 7.6.5
cupti 9.0.176
cycler 0.10.0
decorator 4.4.2
easydict 1.10
enum34 1.1.6
funcsigs 1.0.2
futures 3.3.0
gast 0.5.3
grpcio 1.27.2
intel-openmp 2020.2
kiwisolver 1.1.0
libffi 3.3
libgcc-ng 11.2.0
libgfortran-ng 7.3.0
libprotobuf 3.11.2
libstdcxx-ng 11.2.0
markdown 3.1.1
matplotlib 2.2.5
mkl 2020.2
mkl-service 2.3.0
mkl_fft 1.0.15
mkl_random 1.1.0
mock 3.0.5
ncurses 6.3
networkx 2.2
numpy 1.16.6
numpy-base 1.16.6
opencv-python 4.0.0.21
openssl 1.1.1q
Pillow 6.2.2
pip 19.3.1
protobuf 3.11.2
pyparsing 2.4.7
python 2.7.12
python-dateutil 2.8.2
pytz 2022.5
PyWavelets 1.0.3
readline 8.1.2
scikit-image 0.14.5
scipy 1.2.3
setuptools 44.0.0
six 1.16.0
sqlite 3.39.3
subprocess32 3.5.4
tensorboard 1.10.0
tensorflow 1.10.0
tensorflow-base 1.10.0
tensorflow-gpu 1.10.0
termcolor 1.1.0
tk 8.6.12
werkzeug 1.0.1
wheel 0.37.1
zlib 1.2.12

Your work has great impacts on my thoughts about defocus deblurring, thank you for doing this amazing work!
However, the python scripts in this repository are all evaluation codes for pretrained models. Is it possible for you to provide the training code?
Thanks a lot!