Skip to content

Latest commit

 

History

History
48 lines (41 loc) · 1.58 KB

INSTALL.md

File metadata and controls

48 lines (41 loc) · 1.58 KB
Raw

Setup

Libraries

Guaranteed to work on Ubuntu 16.04 with Python 3.7+, CUDA 10.1 and also:

pip3 install \
    torch==1.5.1 \
    torchvision==0.6.1

# This is my FULL environment, so likely not all these packages are actually required
pip3 install \
    tensorboard==2.2.2 \
    pandas==1.0.5 \
    torchsummary==1.5.1 \
    thop==0.0.31-2005241907 \
    tqdm==4.47.0 \
    scipy==1.5.1 \
    opencv-python==4.3.0.36 \
    imageio==2.9.0 \
    imageio-ffmpeg==0.4.2 \
    pymemcache==3.2.0 \
    python-memcached==1.59 \
    matplotlib==3.2.2 \
    seaborn==0.10.1 \
    huepy==1.2.1 \
    albumentations==0.4.6 \
    imgaug==0.4.0 \
    PyYAML==5.3.1 \
    munch==2.5.0 \
    pickle5==0.0.11 \
    jpeg4py==0.1.4 \
    face-alignment==1.1.1 \
    mxnet-cu101mkl==1.6.0.post0 \
    scikit-learn==0.23.2

pip3 install git+https://github.com/DmitryUlyanov/yamlenv.git
pip3 install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cuda_ext" git+https://github.com/NVIDIA/apex.git
pip3 install git+https://github.com/dniku/insightface.git@763060b2

Repository

  1. git submodule update --init
  2. Download the background segmentation model's checkpoint and put it to 'utils/Graphonomy/data/model'.
  3. Download the two VGG weight files (first, second) for perceptual losses and put them to 'criterions/common/'.