Style transfer, deep learning, feature transform

I am receiving this error: RuntimeError: the number of sizes provided must be greater or equal to the number of dimensions in the tensor at /opt/conda/conda-bld/pytorch_1501972792122/work/pytorch-0.1.12/torch/lib/THC/generic/THCTensor.c:299

Hi I'm a bit new to Python and have trouble understanding the messages I get when running "converter.py" in Terminal:

usage: cp [-R [-H | -L | -P]] [-fi | -n] [-apvXc] source_file target_file cp [-R [-H | -L | -P]] [-fi | -n] [-apvXc] source_file ... target_directory

What am I supposed to do next? These don't seem to be standard Python asks and I couldn't find a user guide on how to use this script. Forgive me if I'm missing something obvious

Thanks for the great code. When I run the algorithm with my own high-resolution images (655 * 1280), I find that when using scipy.sparse.linalg.spsolve with scikit-umfpack as solver, it requires too much memory (larger than 128GB). After some investigations, I found the problem might be OS dependent. However, I actually followed the instructions: my OS is Ubuntu 16.04, also the same CUDA and python version.

I wonder if anyone struggles at the same issue with me, and if there is any other solver. Thanks.

What am I doing wrong? The simple demo with global style works, but trying with label maps I get an error.

Picture of the images and the visualized label maps:

I run this command:

python demo.py \
--content_image_path images/custom2/content1.png \
--content_seg_path images/custom2/content1.label/label.png \
--style_image_path images/custom2/style1.png \
--style_seg_path images/custom2/style1.label/label.png \
--output_image_path results/example2.png

Output and error:

Elapsed time in stylization: 0.417996
Traceback (most recent call last):
  File "demo.py", line 43, in <module>
    cuda=args.cuda,
  File "/home/ubuntu/.fast-photo-style/process_stylization.py", line 62, in stylization
    stylized_img = p_wct.transform(cont_img, styl_img, cont_seg, styl_seg)
  File "/home/ubuntu/.fast-photo-style/photo_wct.py", line 35, in transform
    csF4 = self.__feature_wct(cF4, sF4, cont_seg, styl_seg)
  File "/home/ubuntu/.fast-photo-style/photo_wct.py", line 88, in __feature_wct
    cont_mask = np.where(t_cont_seg.reshape(t_cont_seg.shape[0] * t_cont_seg.shape[1]) == l)
ValueError: total size of new array must be unchanged

THCudaCheck FAIL file=/opt/conda/conda-bld/pytorch_1501969512886/work/pytorch-0.1.12/torch/lib/THC/generic/THCStorage.cu line=66 error=2 : out of memory Traceback (most recent call last): File "demo.py", line 68, in stylized_img = p_wct.transform(cont_img, styl_img, cont_seg, styl_seg) File "/home/boss/FastPhotoStyle-master/photo_wct.py", line 36, in transform sF4,sF3,sF2,sF1 = self.e4.forward_multiple(styl_img) File "/home/boss/FastPhotoStyle-master/models.py", line 393, in forward_multiple out1 = self.conv3(out1) File "/home/boss/anaconda2/envs/NVIDIA/lib/python3.5/site-packages/torch/nn/modules/module.py", line 206, in call result = self.forward(*input, **kwargs) File "/home/boss/anaconda2/envs/NVIDIA/lib/python3.5/site-packages/torch/nn/modules/conv.py", line 237, in forward self.padding, self.dilation, self.groups) File "/home/boss/anaconda2/envs/NVIDIA/lib/python3.5/site-packages/torch/nn/functional.py", line 40, in conv2d return f(input, weight, bias) RuntimeError: cuda runtime error (2) : out of memory at /opt/conda/conda-bld/pytorch_1501969512886/work/pytorch-0.1.12/torch/lib/THC/generic/THCStorage.cu:66

Hi, I tested your code by running demo.sh with a K40m GPU, but my CUDA version is 8.0 (not 9.1). The total time is about 145s, more than 10 times slower than the reported time in the paper (11.39s for 1K image size). Besides a better GPU (Titan XP), I wonder whether the new CUDA is the key for the high performance. Thanks.

Hi, I am running FastPhotoStyle code on Windows 10 and using Python 3.7, CUDA 10.0 and cuda 9.1. Although I made the change that was suggested to upgrade the version of Python from string to Byte, I am still getting the same error. Can you please suggest a fix for this issue.

Resize image: (803,538)->(803,538) Resize image: (960,540)->(960,540) Elapsed time in stylization: 2.325060 Elapsed time in propagation: 83.987388 Elapsed time in post processing: 0.015629 Traceback (most recent call last): File "demo.py", line 47, in no_post=args.no_post File "D:\TrainImages\FastPhotoStyle-master\process_stylization.py", line 135, in stylization out_img = smooth_filter(out_img, cont_pilimg, f_radius=15, f_edge=1e-1) File "D:\TrainImages\FastPhotoStyle-master\smooth_filter.py", line 402, in smooth_filter best_ = smooth_local_affine(output_, input_, 1e-7, 3, H, W, f_radius, f_edge) File "D:\TrainImages\FastPhotoStyle-master\smooth_filter.py", line 333, in smooth_local_affine program = Program(src.encode('utf-8'), 'best_local_affine_kernel.cu'.encode('utf-8')) File "C:\Users\SD\Anaconda3\lib\site-packages\pynvrtc\compiler.py", line 49, in init self._interface = NVRTCInterface(lib_name) File "C:\Users\SD\Anaconda3\lib\site-packages\pynvrtc\interface.py", line 87, in init self._load_nvrtc_lib(lib_path) File "C:\Users\SD\Anaconda3\lib\site-packages\pynvrtc\interface.py", line 109, in _load_nvrtc_lib self.lib = cdll.LoadLibrary(name) File "C:\Users\SD\Anaconda3\lib\ctypes_init.py", line 434, in LoadLibrary return self.dlltype(name) File "C:\Users\SD\Anaconda3\lib\ctypes_init.py", line 356, in init self._handle = _dlopen(self._name, mode) OSError: [WinError 126] The specified module could not be found

This addresses issue #55. One expects a 1 channel mask (resize based only on height and width) but you force this to become a 3 channel mask upon loading (mode="RGB"). Now we correctly get a 1 channel 8 bit mask.

Command:

pip install cupy

Result:

ERROR: Complete output from command python setup.py egg_info:
    ERROR: Options: {'package_name': 'cupy', 'long_description': None, 'wheel_libs': [], 'wheel_includes': [], 'no_rpath': False, 'profile': False, 'linetrace': False, 'annotate': False, 'no_cuda': False}
    
    -------- Configuring Module: cuda --------
    Microsoft Visual C++ 14.0 is required. Get it with "Microsoft Visual C++ Build Tools": https://visualstudio.microsoft.com/downloads/
    
    ************************************************************
    * CuPy Configuration Summary                               *
    ************************************************************
    
    Build Environment:
      Include directories: ['C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.0\\include', 'C:\\Program Files\\NVIDIA Corporation\\NvToolsExt\\include']
      Library directories: ['C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.0\\bin', 'C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.0\\lib\\x64', 'C:\\Program Files\\NVIDIA Corporation\\NvToolsExt\\lib\\x64']
      nvcc command       : ['C:\\Program Files\\NVIDIA GPU Computing Toolkit\\CUDA\\v10.0\\bin/nvcc.exe']
    
    Environment Variables:
      CFLAGS          : (none)
      LDFLAGS         : (none)
      LIBRARY_PATH    : (none)
      CUDA_PATH       : C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v10.0
      NVTOOLSEXT_PATH : C:\Program Files\NVIDIA Corporation\NvToolsExt\
      NVCC            : (none)
    
    Modules:
      cuda      : No
        -> Include files not found: ['cublas_v2.h', 'cuda.h', 'cuda_profiler_api.h', 'cuda_runtime.h', 'cufft.h', 'curand.h', 'cusparse.h', 'nvrtc.h']
        -> Check your CFLAGS environment variable.
    
    ERROR: CUDA could not be found on your system.
    Please refer to the Installation Guide for details:
    https://docs-cupy.chainer.org/en/stable/install.html
    
    ************************************************************
    
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "C:\Users\FLAMES~1\AppData\Local\Temp\pip-install-ghow8_pv\cupy\setup.py", line 120, in <module>
        ext_modules = cupy_setup_build.get_ext_modules()
      File "C:\Users\FLAMES~1\AppData\Local\Temp\pip-install-ghow8_pv\cupy\cupy_setup_build.py", line 632, in get_ext_modules
        extensions = make_extensions(arg_options, compiler, use_cython)
      File "C:\Users\FLAMES~1\AppData\Local\Temp\pip-install-ghow8_pv\cupy\cupy_setup_build.py", line 387, in make_extensions
        raise Exception('Your CUDA environment is invalid. '
    Exception: Your CUDA environment is invalid. Please check above error log.
    ----------------------------------------
ERROR: Command "python setup.py egg_info" failed with error code 1 in C:\Users\FLAMES~1\AppData\Local\Temp\pip-install-ghow8_pv\cupy\

Within the paper, I can only see smoothing mentioned once. However in the implementation smoothing is performed twice in photo_smooth.py and smooth_filter.py.

Am I misunderstanding the paper/implementation regarding the second smoothing technique, or is this an addition made? If so, can you explain why this was added?

This PR:

• Convert torch models to pytorch models (listed in TODOs in the origin code) and converter.py shows how it was done. The pytorch model leaves in the submodule PhotoWCTModels which makes it easier to download from a server as https://github.com/NVIDIA/FastPhotoStyle/issues/15 suggested.

• The models are refactored into less and clear classes. The layers are named according to the origin paper.

• Fix a bug in Propagator. It fails to process images with alpha channels because it does not open them with RGB mode.

• CPU support for PhotoWCT. PhotoWCT can work in CPU mode without using .cuda(). This could make it ~10x slower (not too slow yet) but more friendly for those without GPUs or GPUs with less memory as https://github.com/NVIDIA/FastPhotoStyle/issues/17

I'm sorry that some codes in photo_wct.py are changed by the (PEP8) code formatter, so not too much of them are actually modified.

Current code are tested. They can work well as before.

Hello, thanks for your great work I face not found the dataset when run with demo_example3.sh. Could you guide me where to found this seg.dataset folder. https://github.com/CSAILVision/semantic-segmentation-pytorch looks dont have this folder .

Thanks!

The problem The code was comparing booleans using the operator '==', where in Python the indicated is to use the operator 'is', otherwise we would fall into a literal comparison pitfall. This pitfall was detected using Pylint and generated the following message error code and message: Pylint code: C0121

Comparison 'styl_seg.size == False' should be 'styl_seg.size is False' if checking for the singleton value False, or 'not styl_seg.size' if testing for falsiness

Solution Removed the '==' operator and changed it to 'is'

Hi! The docker image fails to build. We get this issue during step 12:

Step 12/16 : RUN conda install -y -c anaconda pip ---> Running in 15c22eca2c92 Collecting package metadata (repodata.json): ...working... done Solving environment: ...working... The environment is inconsistent, please check the package plan carefully The following packages are causing the inconsistency:

• https://repo.continuum.io/pkgs/main/linux-64/conda-verify-2.0.0-py36h98955d8_0.tar.bz2/linux-64::conda-verify==2.0.0=py36h98955d8_0
• https://repo.continuum.io/pkgs/main/linux-64/dask-core-0.15.3-py36h10e6167_0.tar.bz2/linux-64::dask-core==0.15.3=py36h10e6167_0
• https://repo.continuum.io/pkgs/main/linux-64/cython-0.26.1-py36h21c49d0_0.tar.bz2/linux-64::cython==0.26.1=py36h21c49d0_0
• https://repo.continuum.io/pkgs/main/linux-64/dask-0.15.3-py36hdc2c8aa_0.tar.bz2/linux-64::dask==0.15.3=py36hdc2c8aa_0
• https://repo.continuum.io/pkgs/main/linux-64/snowballstemmer-1.2.1-py36h6febd40_0.tar.bz2/linux-64::snowballstemmer==1.2.1=py36h6febd40_0
• https://repo.continuum.io/pkgs/main/linux-64/greenlet-0.4.12-py36h2d503a6_0.tar.bz2/linux-64::greenlet==0.4.12=py36h2d503a6_0
• https://repo.continuum.io/pkgs/main/linux-64/ipython_genutils-0.2.0-py36hb52b0d5_0.tar.bz2/linux-64::ipython_genutils==0.2.0=py36hb52b0d5_0
• https://repo.continuum.io/pkgs/main/linux-64/cryptography-2.0.3-py36ha225213_1.tar.bz2/linux-64::cryptography==2.0.3=py36ha225213_1
• https://repo.continuum.io/pkgs/main/linux-64/xlrd-1.1.0-py36h1db9f0c_1.tar.bz2/linux-64::xlrd==1.1.0=py36h1db9f0c_1
• https://repo.continuum.io/pkgs/main/linux-64/pep8-1.7.0-py36h26ade29_0.tar.bz2/linux-64::pep8==1.7.0=py36h26ade29_0
• https://repo.continuum.io/pkgs/main/linux-64/astroid-1.5.3-py36hbdb9df2_0.tar.bz2/linux-64::astroid==1.5.3=py36hbdb9df2_0
• https://repo.continuum.io/pkgs/main/linux-64/contextlib2-0.5.5-py36h6c84a62_0.tar.bz2/linux-64::contextlib2==0.5.5=py36h6c84a62_0
• https://repo.continuum.io/pkgs/main/linux-64/patsy-0.4.1-py36ha3be15e_0.tar.bz2/linux-64::patsy==0.4.1=py36ha3be15e_0
• https://repo.continuum.io/pkgs/main/linux-64/h5py-2.7.0-py36he81ebca_1.tar.bz2/linux-64::h5py==2.7.0=py36he81ebca_1
• https://repo.continuum.io/pkgs/main/linux-64/html5lib-0.999999999-py36h2cfc398_0.tar.bz2/linux-64::html5lib==0.999999999=py36h2cfc398_0
• https://repo.continuum.io/pkgs/main/linux-64/astropy-2.0.2-py36ha51211e_4.tar.bz2/linux-64::astropy==2.0.2=py36ha51211e_4
• https://repo.continuum.io/pkgs/main/linux-64/lazy-object-proxy-1.3.1-py36h10fcdad_0.tar.bz2/linux-64::lazy-object-proxy==1.3.1=py36h10fcdad_0
• https://repo.continuum.io/pkgs/main/linux-64/jupyter_client-5.1.0-py36h614e9ea_0.tar.bz2/linux-64::jupyter_client==5.1.0=py36h614e9ea_0
• https://repo.continuum.io/pkgs/main/linux-64/filelock-2.0.12-py36hacfa1f5_0.tar.bz2/linux-64::filelock==2.0.12=py36hacfa1f5_0
• https://repo.continuum.io/pkgs/main/linux-64/qtawesome-0.4.4-py36h609ed8c_0.tar.bz2/linux-64::qtawesome==0.4.4=py36h609ed8c_0
• https://repo.continuum.io/pkgs/main/linux-64/mpmath-0.19-py36h8cc018b_2.tar.bz2/linux-64::mpmath==0.19=py36h8cc018b_2
• https://repo.continuum.io/pkgs/main/linux-64/bkcharts-0.2-py36h735825a_0.tar.bz2/linux-64::bkcharts==0.2=py36h735825a_0
• https://repo.continuum.io/pkgs/main/linux-64/certifi-2017.7.27.1-py36h8b7b77e_0.tar.bz2/linux-64::certifi==2017.7.27.1=py36h8b7b77e_0
• https://repo.continuum.io/pkgs/main/linux-64/ipywidgets-7.0.0-py36h7b55c3a_0.tar.bz2/linux-64::ipywidgets==7.0.0=py36h7b55c3a_0
• https://repo.continuum.io/pkgs/main/linux-64/click-6.7-py36h5253387_0.tar.bz2/linux-64::click==6.7=py36h5253387_0
• https://repo.continuum.io/pkgs/main/linux-64/docutils-0.14-py36hb0f60f5_0.tar.bz2/linux-64::docutils==0.14=py36hb0f60f5_0
• https://repo.continuum.io/pkgs/main/linux-64/tblib-1.3.2-py36h34cf8b6_0.tar.bz2/linux-64::tblib==1.3.2=py36h34cf8b6_0
• https://repo.continuum.io/pkgs/main/linux-64/singledispatch-3.4.0.3-py36h7a266c3_0.tar.bz2/linux-64::singledispatch==3.4.0.3=py36h7a266c3_0
• https://repo.continuum.io/pkgs/main/linux-64/asn1crypto-0.22.0-py36h265ca7c_1.tar.bz2/linux-64::asn1crypto==0.22.0=py36h265ca7c_1
• https://repo.continuum.io/pkgs/main/linux-64/jedi-0.10.2-py36h552def0_0.tar.bz2/linux-64::jedi==0.10.2=py36h552def0_0
• https://repo.continuum.io/pkgs/main/linux-64/distributed-1.19.1-py36h25f3894_0.tar.bz2/linux-64::distributed==1.19.1=py36h25f3894_0
• https://repo.continuum.io/pkgs/main/linux-64/pycparser-2.18-py36hf9f622e_1.tar.bz2/linux-64::pycparser==2.18=py36hf9f622e_1
• https://repo.continuum.io/pkgs/main/linux-64/pyodbc-4.0.17-py36h999153c_0.tar.bz2/linux-64::pyodbc==4.0.17=py36h999153c_0
• https://repo.continuum.io/pkgs/main/linux-64/qt-5.6.2-h974d657_12.tar.bz2/linux-64::qt==5.6.2=h974d657_12
• https://repo.continuum.io/pkgs/main/linux-64/openssl-1.0.2l-h077ae2c_5.tar.bz2/linux-64::openssl==1.0.2l=h077ae2c_5
• https://repo.continuum.io/pkgs/main/linux-64/beautifulsoup4-4.6.0-py36h49b8c8c_1.tar.bz2/linux-64::beautifulsoup4==4.6.0=py36h49b8c8c_1
• https://repo.continuum.io/pkgs/main/linux-64/llvmlite-0.20.0-py36_0.tar.bz2/linux-64::llvmlite==0.20.0=py36_0
• https://repo.continuum.io/pkgs/main/linux-64/scikit-image-0.13.0-py36had3c07a_1.tar.bz2/linux-64::scikit-image==0.13.0=py36had3c07a_1
• https://repo.continuum.io/pkgs/main/linux-64/ipykernel-4.6.1-py36hbf841aa_0.tar.bz2/linux-64::ipykernel==4.6.1=py36hbf841aa_0
• https://repo.continuum.io/pkgs/main/linux-64/nltk-3.2.4-py36h1a0979f_0.tar.bz2/linux-64::nltk==3.2.4=py36h1a0979f_0
• https://repo.continuum.io/pkgs/main/linux-64/jupyterlab_launcher-0.4.0-py36h4d8058d_0.tar.bz2/linux-64::jupyterlab_launcher==0.4.0=py36h4d8058d_0
• https://repo.continuum.io/pkgs/main/linux-64/mistune-0.7.4-py36hbab8784_0.tar.bz2/linux-64::mistune==0.7.4=py36hbab8784_0
• https://repo.continuum.io/pkgs/main/linux-64/_ipyw_jlab_nb_ext_conf-0.1.0-py36he11e457_0.tar.bz2/linux-64::_ipyw_jlab_nb_ext_conf==0.1.0=py36he11e457_0
• https://repo.continuum.io/pkgs/main/linux-64/wheel-0.29.0-py36he7f4e38_1.tar.bz2/linux-64::wheel==0.29.0=py36he7f4e38_1
• https://repo.continuum.io/pkgs/main/linux-64/bitarray-0.8.1-py36h5834eb8_0.tar.bz2/linux-64::bitarray==0.8.1=py36h5834eb8_0
• https://repo.continuum.io/pkgs/main/linux-64/ipython-6.1.0-py36hc72a948_1.tar.bz2/linux-64::ipython==6.1.0=py36hc72a948_1
• https://repo.continuum.io/pkgs/main/linux-64/pywavelets-0.5.2-py36he602eb0_0.tar.bz2/linux-64::pywavelets==0.5.2=py36he602eb0_0
• https://repo.continuum.io/pkgs/main/linux-64/six-1.11.0-py36h372c433_1.tar.bz2/linux-64::six==1.11.0=py36h372c433_1
• https://repo.continuum.io/pkgs/main/linux-64/bottleneck-1.2.1-py36haac1ea0_0.tar.bz2/linux-64::bottleneck==1.2.1=py36haac1ea0_0
• https://repo.continuum.io/pkgs/main/linux-64/isort-4.2.15-py36had401c0_0.tar.bz2/linux-64::isort==4.2.15=py36had401c0_0
• https://repo.continuum.io/pkgs/main/linux-64/gmpy2-2.0.8-py36h55090d7_1.tar.bz2/linux-64::gmpy2==2.0.8=py36h55090d7_1
• https://repo.continuum.io/pkgs/main/linux-64/markupsafe-1.0-py36hd9260cd_1.tar.bz2/linux-64::markupsafe==1.0=py36hd9260cd_1
• https://repo.continuum.io/pkgs/main/linux-64/get_terminal_size-1.0.0-haa9412d_0.tar.bz2/linux-64::get_terminal_size==1.0.0=haa9412d_0
• https://repo.continuum.io/pkgs/main/linux-64/sympy-1.1.1-py36hc6d1c1c_0.tar.bz2/linux-64::sympy==1.1.1=py36hc6d1c1c_0
• https://repo.continuum.io/pkgs/main/linux-64/odo-0.5.1-py36h90ed295_0.tar.bz2/linux-64::odo==0.5.1=py36h90ed295_0
• https://repo.continuum.io/pkgs/main/linux-64/msgpack-python-0.4.8-py36hec4c5d1_0.tar.bz2/linux-64::msgpack-python==0.4.8=py36hec4c5d1_0
• https://repo.continuum.io/pkgs/main/linux-64/olefile-0.44-py36h79f9f78_0.tar.bz2/linux-64::olefile==0.44=py36h79f9f78_0
• https://repo.continuum.io/pkgs/main/linux-64/tornado-4.5.2-py36h1283b2a_0.tar.bz2/linux-64::tornado==4.5.2=py36h1283b2a_0
• https://repo.continuum.io/pkgs/main/linux-64/sortedcollections-0.5.3-py36h3c761f9_0.tar.bz2/linux-64::sortedcollections==0.5.3=py36h3c761f9_0
• https://repo.continuum.io/pkgs/main/linux-64/flask-cors-3.0.3-py36h2d857d3_0.tar.bz2/linux-64::flask-cors==3.0.3=py36h2d857d3_0
• https://repo.continuum.io/pkgs/main/linux-64/pysocks-1.6.7-py36hd97a5b1_1.tar.bz2/linux-64::pysocks==1.6.7=py36hd97a5b1_1
• https://repo.continuum.io/pkgs/main/linux-64/sphinxcontrib-1.0-py36h6d0f590_1.tar.bz2/linux-64::sphinxcontrib==1.0=py36h6d0f590_1
• https://repo.continuum.io/pkgs/main/linux-64/pkginfo-1.4.1-py36h215d178_1.tar.bz2/linux-64::pkginfo==1.4.1=py36h215d178_1
• https://repo.continuum.io/pkgs/main/linux-64/sphinx-1.6.3-py36he5f0bdb_0.tar.bz2/linux-64::sphinx==1.6.3=py36he5f0bdb_0
• https://repo.continuum.io/pkgs/main/linux-64/mccabe-0.6.1-py36h5ad9710_1.tar.bz2/linux-64::mccabe==0.6.1=py36h5ad9710_1
• https://repo.continuum.io/pkgs/main/linux-64/simplegeneric-0.8.1-py36h2cb9092_0.tar.bz2/linux-64::simplegeneric==0.8.1=py36h2cb9092_0
• https://repo.continuum.io/pkgs/main/linux-64/itsdangerous-0.24-py36h93cc618_1.tar.bz2/linux-64::itsdangerous==0.24=py36h93cc618_1
• https://repo.continuum.io/pkgs/main/linux-64/xlsxwriter-1.0.2-py36h3de1aca_0.tar.bz2/linux-64::xlsxwriter==1.0.2=py36h3de1aca_0
• https://repo.continuum.io/pkgs/main/linux-64/pandas-0.20.3-py36h842e28d_2.tar.bz2/linux-64::pandas==0.20.3=py36h842e28d_2
• https://repo.continuum.io/pkgs/main/linux-64/requests-2.18.4-py36he2e5f8d_1.tar.bz2/linux-64::requests==2.18.4=py36he2e5f8d_1
• https://repo.continuum.io/pkgs/main/linux-64/pytest-3.2.1-py36h11ad3bb_1.tar.bz2/linux-64::pytest==3.2.1=py36h11ad3bb_1
• https://repo.continuum.io/pkgs/main/linux-64/werkzeug-0.12.2-py36hc703753_0.tar.bz2/linux-64::werkzeug==0.12.2=py36hc703753_0
• https://repo.continuum.io/pkgs/main/linux-64/jupyter_core-4.3.0-py36h357a921_0.tar.bz2/linux-64::jupyter_core==4.3.0=py36h357a921_0
• https://repo.continuum.io/pkgs/main/linux-64/pixman-0.34.0-h83dc358_2.tar.bz2/linux-64::pixman==0.34.0=h83dc358_2
• https://repo.continuum.io/pkgs/main/linux-64/qtconsole-4.3.1-py36h8f73b5b_0.tar.bz2/linux-64::qtconsole==4.3.1=py36h8f73b5b_0
• https://repo.continuum.io/pkgs/main/linux-64/datashape-0.5.4-py36h3ad6b5c_0.tar.bz2/linux-64::datashape==0.5.4=py36h3ad6b5c_0
• https://repo.continuum.io/pkgs/main/linux-64/nbconvert-5.3.1-py36hb41ffb7_0.tar.bz2/linux-64::nbconvert==5.3.1=py36hb41ffb7_0
• https://repo.continuum.io/pkgs/main/linux-64/sqlalchemy-1.1.13-py36hfb5efd7_0.tar.bz2/linux-64::sqlalchemy==1.1.13=py36hfb5efd7_0
• https://repo.continuum.io/pkgs/main/linux-64/pylint-1.7.4-py36hb9d4533_0.tar.bz2/linux-64::pylint==1.7.4=py36hb9d4533_0
• https://repo.continuum.io/pkgs/main/linux-64/bokeh-0.12.10-py36hbb0e44a_0.tar.bz2/linux-64::bokeh==0.12.10=py36hbb0e44a_0
• https://repo.continuum.io/pkgs/main/linux-64/imageio-2.2.0-py36he555465_0.tar.bz2/linux-64::imageio==2.2.0=py36he555465_0
• https://repo.continuum.io/pkgs/main/linux-64/chardet-3.0.4-py36h0f667ec_1.tar.bz2/linux-64::chardet==3.0.4=py36h0f667ec_1
• https://repo.continuum.io/pkgs/main/linux-64/spyder-3.2.4-py36hbe6152b_0.tar.bz2/linux-64::spyder==3.2.4=py36hbe6152b_0
• https://repo.continuum.io/pkgs/main/linux-64/testpath-0.3.1-py36h8cadb63_0.tar.bz2/linux-64::testpath==0.3.1=py36h8cadb63_0
• https://repo.continuum.io/pkgs/main/linux-64/flask-0.12.2-py36hb24657c_0.tar.bz2/linux-64::flask==0.12.2=py36hb24657c_0
• https://repo.continuum.io/pkgs/main/linux-64/jdcal-1.3-py36h4c697fb_0.tar.bz2/linux-64::jdcal==1.3=py36h4c697fb_0
• https://repo.continuum.io/pkgs/main/linux-64/anaconda-client-1.6.5-py36h19c0dcd_0.tar.bz2/linux-64::anaconda-client==1.6.5=py36h19c0dcd_0
• https://repo.continuum.io/pkgs/main/linux-64/pandocfilters-1.4.2-py36ha6701b7_1.tar.bz2/linux-64::pandocfilters==1.4.2=py36ha6701b7_1
• https://repo.continuum.io/pkgs/main/linux-64/pygments-2.2.0-py36h0d3125c_0.tar.bz2/linux-64::pygments==2.2.0=py36h0d3125c_0
• https://repo.continuum.io/pkgs/main/linux-64/webencodings-0.5.1-py36h800622e_1.tar.bz2/linux-64::webencodings==0.5.1=py36h800622e_1
• https://repo.continuum.io/pkgs/main/linux-64/qtpy-1.3.1-py36h3691cc8_0.tar.bz2/linux-64::qtpy==1.3.1=py36h3691cc8_0
• https://repo.continuum.io/pkgs/main/linux-64/pexpect-4.2.1-py36h3b9d41b_0.tar.bz2/linux-64::pexpect==4.2.1=py36h3b9d41b_0
• https://repo.continuum.io/pkgs/main/linux-64/pyyaml-3.12-py36hafb9ca4_1.tar.bz2/linux-64::pyyaml==3.12=py36hafb9ca4_1
• https://repo.continuum.io/pkgs/main/linux-64/python-3.6.3-hc9025b9_1.tar.bz2/linux-64::python==3.6.3=hc9025b9_1
• https://repo.continuum.io/pkgs/main/linux-64/terminado-0.6-py36ha25a19f_0.tar.bz2/linux-64::terminado==0.6=py36ha25a19f_0
• https://repo.continuum.io/pkgs/main/linux-64/jupyter-1.0.0-py36h9896ce5_0.tar.bz2/linux-64::jupyter==1.0.0=py36h9896ce5_0
• https://repo.continuum.io/pkgs/main/linux-64/et_xmlfile-1.0.1-py36hd6bccc3_0.tar.bz2/linux-64::et_xmlfile==1.0.1=py36hd6bccc3_0
• https://repo.continuum.io/pkgs/main/linux-64/notebook-5.0.0-py36h0b20546_2.tar.bz2/linux-64::notebook==5.0.0=py36h0b20546_2
• https://repo.continuum.io/pkgs/main/linux-64/ptyprocess-0.5.2-py36h69acd42_0.tar.bz2/linux-64::ptyprocess==0.5.2=py36h69acd42_0
• https://repo.continuum.io/pkgs/main/linux-64/pytz-2017.2-py36hc2ccc2a_1.tar.bz2/linux-64::pytz==2017.2=py36hc2ccc2a_1
• https://repo.continuum.io/pkgs/main/linux-64/cycler-0.10.0-py36h93f1223_0.tar.bz2/linux-64::cycler==0.10.0=py36h93f1223_0
• https://repo.continuum.io/pkgs/main/linux-64/sphinxcontrib-websupport-1.0.1-py36hb5cb234_1.tar.bz2/linux-64::sphinxcontrib-websupport==1.0.1=py36hb5cb234_1
• https://repo.continuum.io/pkgs/main/linux-64/pyqt-5.6.0-py36h0386399_5.tar.bz2/linux-64::pyqt==5.6.0=py36h0386399_5
• https://repo.continuum.io/pkgs/main/linux-64/cloudpickle-0.4.0-py36h30f8c20_0.tar.bz2/linux-64::cloudpickle==0.4.0=py36h30f8c20_0
• https://repo.continuum.io/pkgs/main/linux-64/pyflakes-1.6.0-py36h7bd6a15_0.tar.bz2/linux-64::pyflakes==1.6.0=py36h7bd6a15_0
• https://repo.continuum.io/pkgs/main/linux-64/numpydoc-0.7.0-py36h18f165f_0.tar.bz2/linux-64::numpydoc==0.7.0=py36h18f165f_0
• https://repo.continuum.io/pkgs/main/linux-64/pickleshare-0.7.4-py36h63277f8_0.tar.bz2/linux-64::pickleshare==0.7.4=py36h63277f8_0
• https://repo.continuum.io/pkgs/main/linux-64/wcwidth-0.1.7-py36hdf4376a_0.tar.bz2/linux-64::wcwidth==0.1.7=py36hdf4376a_0
• https://repo.continuum.io/pkgs/main/linux-64/sip-4.18.1-py36h51ed4ed_2.tar.bz2/linux-64::sip==4.18.1=py36h51ed4ed_2
• https://repo.continuum.io/pkgs/main/linux-64/navigator-updater-0.1.0-py36h14770f7_0.tar.bz2/linux-64::navigator-updater==0.1.0=py36h14770f7_0
• https://repo.continuum.io/pkgs/main/linux-64/babel-2.5.0-py36h7d14adf_0.tar.bz2/linux-64::babel==2.5.0=py36h7d14adf_0
• https://repo.continuum.io/pkgs/main/linux-64/nbformat-4.4.0-py36h31c9010_0.tar.bz2/linux-64::nbformat==4.4.0=py36h31c9010_0
• https://repo.continuum.io/pkgs/main/linux-64/zict-0.1.3-py36h3a3bf81_0.tar.bz2/linux-64::zict==0.1.3=py36h3a3bf81_0
• https://repo.continuum.io/pkgs/main/linux-64/statsmodels-0.8.0-py36h8533d0b_0.tar.bz2/linux-64::statsmodels==0.8.0=py36h8533d0b_0
• https://repo.continuum.io/pkgs/main/linux-64/pycurl-7.43.0-py36h5e72054_3.tar.bz2/linux-64::pycurl==7.43.0=py36h5e72054_3
• https://repo.continuum.io/pkgs/main/linux-64/seaborn-0.8.0-py36h197244f_0.tar.bz2/linux-64::seaborn==0.8.0=py36h197244f_0
• https://repo.continuum.io/pkgs/main/linux-64/pillow-4.2.1-py36h9119f52_0.tar.bz2/linux-64::pillow==4.2.1=py36h9119f52_0
• https://repo.continuum.io/pkgs/main/linux-64/pycrypto-2.6.1-py36h6998063_1.tar.bz2/linux-64::pycrypto==2.6.1=py36h6998063_1
• https://repo.continuum.io/pkgs/main/linux-64/mkl-service-1.1.2-py36h17a0993_4.tar.bz2/linux-64::mkl-service==1.1.2=py36h17a0993_4
• https://repo.continuum.io/pkgs/main/linux-64/jupyterlab-0.27.0-py36h86377d0_2.tar.bz2/linux-64::jupyterlab==0.27.0=py36h86377d0_2
• https://repo.continuum.io/pkgs/main/linux-64/conda-build-3.0.27-py36h940a66d_0.tar.bz2/linux-64::conda-build==3.0.27=py36h940a66d_0
• https://repo.continuum.io/pkgs/main/linux-64/jupyter_console-5.2.0-py36he59e554_1.tar.bz2/linux-64::jupyter_console==5.2.0=py36he59e554_1
• https://repo.continuum.io/pkgs/main/linux-64/numexpr-2.6.2-py36hdd3393f_1.tar.bz2/linux-64::numexpr==2.6.2=py36hdd3393f_1
• https://repo.continuum.io/pkgs/main/linux-64/nose-1.3.7-py36hcdf7029_2.tar.bz2/linux-64::nose==1.3.7=py36hcdf7029_2
• https://repo.continuum.io/pkgs/main/linux-64/wrapt-1.10.11-py36h28b7045_0.tar.bz2/linux-64::wrapt==1.10.11=py36h28b7045_0
• https://repo.continuum.io/pkgs/main/linux-64/xlwt-1.3.0-py36h7b00a1f_0.tar.bz2/linux-64::xlwt==1.3.0=py36h7b00a1f_0
• https://repo.continuum.io/pkgs/main/linux-64/jinja2-2.9.6-py36h489bce4_1.tar.bz2/linux-64::jinja2==2.9.6=py36h489bce4_1
• https://repo.continuum.io/pkgs/main/linux-64/decorator-4.1.2-py36hd076ac8_0.tar.bz2/linux-64::decorator==4.1.2=py36hd076ac8_0
• https://repo.continuum.io/pkgs/main/linux-64/packaging-16.8-py36ha668100_1.tar.bz2/linux-64::packaging==16.8=py36ha668100_1
• https://repo.continuum.io/pkgs/main/linux-64/harfbuzz-1.5.0-h2545bd6_0.tar.bz2/linux-64::harfbuzz==1.5.0=h2545bd6_0
• https://repo.continuum.io/pkgs/main/linux-64/scipy-0.19.1-py36h9976243_3.tar.bz2/linux-64::scipy==0.19.1=py36h9976243_3
• https://repo.continuum.io/pkgs/main/linux-64/numpy-1.13.3-py36ha12f23b_0.tar.bz2/linux-64::numpy==1.13.3=py36ha12f23b_0
• https://repo.continuum.io/pkgs/main/linux-64/typing-3.6.2-py36h7da032a_0.tar.bz2/linux-64::typing==3.6.2=py36h7da032a_0
• https://repo.continuum.io/pkgs/main/linux-64/pango-1.40.11-h8191d47_0.tar.bz2/linux-64::pango==1.40.11=h8191d47_0
• https://repo.continuum.io/pkgs/main/linux-64/entrypoints-0.2.3-py36h1aec115_2.tar.bz2/linux-64::entrypoints==0.2.3=py36h1aec115_2
• https://repo.continuum.io/pkgs/main/linux-64/ruamel_yaml-0.11.14-py36ha2fb22d_2.tar.bz2/linux-64::ruamel_yaml==0.11.14=py36ha2fb22d_2
• https://repo.continuum.io/pkgs/main/linux-64/pytables-3.4.2-py36h3b5282a_2.tar.bz2/linux-64::pytables==3.4.2=py36h3b5282a_2
• https://repo.continuum.io/pkgs/main/linux-64/pyzmq-16.0.2-py36h3b0cf96_2.tar.bz2/linux-64::pyzmq==16.0.2=py36h3b0cf96_2
• https://repo.continuum.io/pkgs/main/linux-64/locket-0.2.0-py36h787c0ad_1.tar.bz2/linux-64::locket==0.2.0=py36h787c0ad_1
• https://repo.continuum.io/pkgs/main/linux-64/toolz-0.8.2-py36h81f2dff_0.tar.bz2/linux-64::toolz==0.8.2=py36h81f2dff_0
• https://repo.continuum.io/pkgs/main/linux-64/anaconda-navigator-1.6.9-py36h11ddaaa_0.tar.bz2/linux-64::anaconda-navigator==1.6.9=py36h11ddaaa_0
• https://repo.continuum.io/pkgs/main/linux-64/heapdict-1.0.0-py36h79797d7_0.tar.bz2/linux-64::heapdict==1.0.0=py36h79797d7_0
• https://repo.continuum.io/pkgs/main/linux-64/setuptools-36.5.0-py36he42e2e1_0.tar.bz2/linux-64::setuptools==36.5.0=py36he42e2e1_0
• https://repo.continuum.io/pkgs/main/linux-64/scikit-learn-0.19.1-py36h7aa7ec6_0.tar.bz2/linux-64::scikit-learn==0.19.1=py36h7aa7ec6_0
• https://repo.continuum.io/pkgs/main/linux-64/curl-7.55.1-hcb0b314_2.tar.bz2/linux-64::curl==7.55.1=hcb0b314_2
• https://repo.continuum.io/pkgs/main/linux-64/multipledispatch-0.4.9-py36h41da3fb_0.tar.bz2/linux-64::multipledispatch==0.4.9=py36h41da3fb_0
• https://repo.continuum.io/pkgs/main/linux-64/lxml-4.1.0-py36h5b66e50_0.tar.bz2/linux-64::lxml==4.1.0=py36h5b66e50_0
• https://repo.continuum.io/pkgs/main/linux-64/bleach-2.0.0-py36h688b259_0.tar.bz2/linux-64::bleach==2.0.0=py36h688b259_0
• https://repo.continuum.io/pkgs/main/linux-64/clyent-1.2.2-py36h7e57e65_1.tar.bz2/linux-64::clyent==1.2.2=py36h7e57e65_1
• https://repo.continuum.io/pkgs/main/linux-64/glob2-0.5-py36h2c1b292_1.tar.bz2/linux-64::glob2==0.5=py36h2c1b292_1
• https://repo.continuum.io/pkgs/main/linux-64/boto-2.48.0-py36h6e4cd66_1.tar.bz2/linux-64::boto==2.48.0=py36h6e4cd66_1
• https://repo.continuum.io/pkgs/main/linux-64/cairo-1.14.10-haa5651f_5.tar.bz2/linux-64::cairo==1.14.10=haa5651f_5
• https://repo.continuum.io/pkgs/main/linux-64/py-1.4.34-py36h0712aa3_1.tar.bz2/linux-64::py==1.4.34=py36h0712aa3_1
• https://repo.continuum.io/pkgs/main/linux-64/pip-9.0.1-py36h8ec8b28_3.tar.bz2/linux-64::pip==9.0.1=py36h8ec8b28_3
• https://repo.continuum.io/pkgs/main/linux-64/fastcache-1.0.2-py36h5b0c431_0.tar.bz2/linux-64::fastcache==1.0.2=py36h5b0c431_0
• https://repo.continuum.io/pkgs/main/linux-64/gevent-1.2.2-py36h2fe25dc_0.tar.bz2/linux-64::gevent==1.2.2=py36h2fe25dc_0
• https://repo.continuum.io/pkgs/main/linux-64/imagesize-0.7.1-py36h52d8127_0.tar.bz2/linux-64::imagesize==0.7.1=py36h52d8127_0
• https://repo.continuum.io/pkgs/main/linux-64/openpyxl-2.4.8-py36h41dd2a8_1.tar.bz2/linux-64::openpyxl==2.4.8=py36h41dd2a8_1
• https://repo.continuum.io/pkgs/main/linux-64/networkx-2.0-py36h7e96fb8_0.tar.bz2/linux-64::networkx==2.0=py36h7e96fb8_0
• https://repo.continuum.io/pkgs/main/linux-64/pathlib2-2.3.0-py36h49efa8e_0.tar.bz2/linux-64::pathlib2==2.3.0=py36h49efa8e_0
• https://repo.continuum.io/pkgs/main/linux-64/blaze-0.11.3-py36h4e06776_0.tar.bz2/linux-64::blaze==0.11.3=py36h4e06776_0
• https://repo.continuum.io/pkgs/main/linux-64/libxcb-1.12-h84ff03f_3.tar.bz2/linux-64::libxcb==1.12=h84ff03f_3
• https://repo.continuum.io/pkgs/main/linux-64/alabaster-0.7.10-py36h306e16b_0.tar.bz2/linux-64::alabaster==0.7.10=py36h306e16b_0
• https://repo.continuum.io/pkgs/main/linux-64/matplotlib-2.1.0-py36hba5de38_0.tar.bz2/linux-64::matplotlib==2.1.0=py36hba5de38_0
• https://repo.continuum.io/pkgs/main/linux-64/pycodestyle-2.3.1-py36hf609f19_0.tar.bz2/linux-64::pycodestyle==2.3.1=py36hf609f19_0
• https://repo.continuum.io/pkgs/main/linux-64/prompt_toolkit-1.0.15-py36h17d85b1_0.tar.bz2/linux-64::prompt_toolkit==1.0.15=py36h17d85b1_0
• https://repo.continuum.io/pkgs/main/linux-64/numba-0.35.0-np113py36_10.tar.bz2/linux-64::numba==0.35.0=np113py36_10
• https://repo.continuum.io/pkgs/main/linux-64/anaconda-5.0.1-py36hd30a520_1.tar.bz2/linux-64::anaconda==5.0.1=py36hd30a520_1
• https://repo.continuum.io/pkgs/main/linux-64/widgetsnbextension-3.0.2-py36hd01bb71_1.tar.bz2/linux-64::widgetsnbextension==3.0.2=py36hd01bb71_1
• https://repo.continuum.io/pkgs/main/linux-64/unicodecsv-0.14.1-py36ha668878_0.tar.bz2/linux-64::unicodecsv==0.14.1=py36ha668878_0
• https://repo.continuum.io/pkgs/main/linux-64/pyparsing-2.2.0-py36hee85983_1.tar.bz2/linux-64::pyparsing==2.2.0=py36hee85983_1
• https://repo.continuum.io/pkgs/main/linux-64/cffi-1.10.0-py36had8d393_1.tar.bz2/linux-64::cffi==1.10.0=py36had8d393_1
• https://repo.continuum.io/pkgs/main/linux-64/pyopenssl-17.2.0-py36h5cc804b_0.tar.bz2/linux-64::pyopenssl==17.2.0=py36h5cc804b_0
• https://repo.continuum.io/pkgs/main/linux-64/rope-0.10.5-py36h1f8c17e_0.tar.bz2/linux-64::rope==0.10.5=py36h1f8c17e_0
• https://repo.continuum.io/pkgs/main/linux-64/cytoolz-0.8.2-py36h708bfd4_0.tar.bz2/linux-64::cytoolz==0.8.2=py36h708bfd4_0
• https://repo.continuum.io/pkgs/main/linux-64/backports-1.0-py36hfa02d7e_1.tar.bz2/linux-64::backports==1.0=py36hfa02d7e_1
• https://repo.continuum.io/pkgs/main/linux-64/urllib3-1.22-py36hbe7ace6_0.tar.bz2/linux-64::urllib3==1.22=py36hbe7ace6_0
• https://repo.continuum.io/pkgs/main/linux-64/python-dateutil-2.6.1-py36h88d3b88_1.tar.bz2/linux-64::python-dateutil==2.6.1=py36h88d3b88_1
• https://repo.continuum.io/pkgs/main/linux-64/sortedcontainers-1.5.7-py36hdf89491_0.tar.bz2/linux-64::sortedcontainers==1.5.7=py36hdf89491_0
• https://repo.continuum.io/pkgs/main/linux-64/ply-3.10-py36hed35086_0.tar.bz2/linux-64::ply==3.10=py36hed35086_0
• https://repo.continuum.io/pkgs/main/linux-64/psutil-5.4.0-py36h84c53db_0.tar.bz2/linux-64::psutil==5.4.0=py36h84c53db_0
• https://repo.continuum.io/pkgs/main/linux-64/jsonschema-2.6.0-py36h006f8b5_0.tar.bz2/linux-64::jsonschema==2.6.0=py36h006f8b5_0
• https://repo.continuum.io/pkgs/main/linux-64/path.py-10.3.1-py36he0c6f6d_0.tar.bz2/linux-64::path.py==10.3.1=py36he0c6f6d_0
• https://repo.continuum.io/pkgs/main/linux-64/traitlets-4.3.2-py36h674d592_0.tar.bz2/linux-64::traitlets==4.3.2=py36h674d592_0
• https://repo.continuum.io/pkgs/main/linux-64/anaconda-project-0.8.0-py36h29abdf5_0.tar.bz2/linux-64::anaconda-project==0.8.0=py36h29abdf5_0
• https://repo.continuum.io/pkgs/main/linux-64/partd-0.3.8-py36h36fd896_0.tar.bz2/linux-64::partd==0.3.8=py36h36fd896_0
• https://repo.continuum.io/pkgs/main/linux-64/backports.shutil_get_terminal_size-1.0.0-py36hfea85ff_2.tar.bz2/linux-64::backports.shutil_get_terminal_size==1.0.0=py36hfea85ff_2
• https://repo.continuum.io/pkgs/main/linux-64/idna-2.6-py36h82fb2a8_1.tar.bz2/linux-64::idna==2.6=py36h82fb2a8_1
• https://repo.continuum.io/pkgs/main/linux-64/colorama-0.3.9-py36h489cec4_0.tar.bz2/linux-64::colorama==0.3.9=py36h489cec4_0
• https://repo.continuum.io/pkgs/main/linux-64/linux-64::ninja==1.8.2=py36h6bb024c_1
• https://repo.continuum.io/pkgs/main/linux-64/linux-64::pycosat==0.6.3=py36h27cfd23_0
• pytorch/noarch::torchvision==0.2.1=py_2
• https://repo.continuum.io/pkgs/main/linux-64/linux-64::conda-package-handling==1.7.2=py36h03888b9_0
• https://repo.continuum.io/pkgs/main/linux-64/linux-64::conda==4.9.2=py36h06a4308_0
• pytorch/linux-64::pytorch==0.4.1=py36_py35_py27__9.0.176_7.1.2_2
• https://repo.continuum.io/pkgs/main/noarch/noarch::tqdm==4.59.0=pyhd3eb1b0_1

Unless I'm doing something wrong, there is a corner case inside def __wct_core, when one of the matrices is basically a vector. When calculating: contentConv = torch.mm(cont_feat, cont_feat.t()).div(cFSize[1] - 1) + iden

cFSize[1] is 1 so there is a division by 0=> and we get a matrix full of NAN which is causing the SVD to fail.

For now, as a w/a inside def __feature_wct I've changed the condtion if cont_mask[0].size <= 0 or styl_mask[0].size <= 0: continue

to

if cont_mask[0].size <= 1 or styl_mask[0].size <= 1: continue

to ignore labels that causing this issue.

Any idea why it happens and what is the best approach to fix it?

Note that demo_example3.sh depends on CSAIL Semantic Segmentation repo, which itself depends on later pytorch version. This example might break -- I added a warning in there.