Official Colab Released!

[cuiaiyu]

We just released an official Colab demo! You can play with the applications that DiOr supports there!

Moreover, if you have troubles with the following issues, you can find the solutions in the colab.

1. Deepfashion dataset access issue. You can consider an alternative dataset DeepFashion-MultiModal, which does not require a zip password. However, you are resonsible to ensure that it is legal for your party to use this dataset by reading their license. The Colab Demo provides all the necessary steps to convert DeepFashion-MultiModal to the appropriate format that can be loaded by this project.
2. GFLA CUDA function Installation Issue. The GFLA installation is not needed for inference. As you will find, the Colab is not trying to install the CUDA functions.
3. Compiling problem with the latest versions of Pytorch. It is the GFLA CUDA functions that require an old version of PyTorch (1.0.0). Since we do not require those CUDA functions for inference, we can use the latest version of PyTorch.
4. How to set up the (img, parse, keypoint) path The Colab shows how to download and place the data from scratch for test set. It should be similar to setting up training data too.

[aravind-h-v]

Hi, there seems to be an error in the installing skimage==0.16.2:

Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/
Collecting scikit-image==0.16.2
Downloading scikit-image-0.16.2.tar.gz (28.9 MB)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 28.9/28.9 MB 22.6 MB/s eta 0:00:00
Preparing metadata (setup.py) ... done
Requirement already satisfied: networkx>=2.0 in /usr/local/lib/python3.10/dist-packages (from scikit-image==0.16.2) (3.1)
Requirement already satisfied: pillow>=4.3.0 in /usr/local/lib/python3.10/dist-packages (from scikit-image==0.16.2) (8.4.0)
Requirement already satisfied: imageio>=2.3.0 in /usr/local/lib/python3.10/dist-packages (from scikit-image==0.16.2) (2.25.1)
Requirement already satisfied: PyWavelets>=0.4.0 in /usr/local/lib/python3.10/dist-packages (from scikit-image==0.16.2) (1.4.1)
Requirement already satisfied: numpy in /usr/local/lib/python3.10/dist-packages (from imageio>=2.3.0->scikit-image==0.16.2) (1.22.4)
Building wheels for collected packages: scikit-image
error: subprocess-exited-with-error

× python setup.py bdist_wheel did not run successfully.
│ exit code: 1
╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip.
Building wheel for scikit-image (setup.py) ... error
ERROR: Failed building wheel for scikit-image
Running setup.py clean for scikit-image
Failed to build scikit-image
ERROR: Could not build wheels for scikit-image, which is required to install pyproject.toml-based projects

[aravind-h-v]

This bunch of commands seem to work though in successfully installing skimage==0.16.2:

%cd /content
!git clone 'https://github.com/scikit-image/scikit-image.git'
%cd /content/scikit-image
!git checkout main
!git pull
!git checkout tags/v0.16.2
!pip install -e .
%cd /content
!ls

[aile1997]

I had the same problem

I solved the first problem with your comment, but there is another problem

[srssethuraman]

This bunch of commands seem to work though in successfully installing skimage==0.16.2:

%cd /content !git clone 'https://github.com/scikit-image/scikit-image.git' %cd /content/scikit-image !git checkout main !git pull !git checkout tags/v0.16.2 !pip install -e . %cd /content !ls

After successful installation of scikit-image from GIT-HUB repo. i got below error. kindly help me to fix this issue

ModuleNotFoundError: No module named 'skimage'