skycrapers · Poikilos · Jun 30, 2022 · Jun 30, 2022 · Jun 30, 2022 · Jun 30, 2022
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,171 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+# TecoGAN-specific files
+/data/ToS3/GT.zip
+/data/ToS3
+/data/Vid4
+/pretrained_models/TecoGAN_BD_iter500000.pth
+/pretrained_models/TecoGAN_2x_BD_REDS_iter500K.pth
+/pretrained_models/TecoGAN_4x_BD_REDS_iter500K.pth
+/pretrained_models/TecoGAN_4x_BD_Vimeo_iter500K.pth
+/pretrained_models/TecoGAN_4x_BI_Vimeo_iter500K.pth
+/results/
diff --git a/README.md b/README.md
@@ -42,9 +42,13 @@ This is a PyTorch reimplementation of **TecoGAN**: **Te**mporally **Co**herent *
 ## Dependencies
 - Ubuntu >= 16.04
 - NVIDIA GPU + CUDA
+  - CUDA Toolkit 11.3 is the latest version with which Pytorch is compatible as of June 2022.
+  - To see which GPU driver versions are compatible with which CUDA Toolkit version, see [CUDA Compatibility](https://docs.nvidia.com/deploy/cuda-compatibility/index.html).
 - Python >= 3.7
 - PyTorch >= 1.4.0
-- Python packages: numpy, matplotlib, opencv-python, pyyaml, lmdb
+- Python packages: See [requirements.txt](requirements.txt)
+  - A higher version of pytorch and torchvision may work, but torch==1.7.1 (torchvision==0.8.2 seems compatible) is the lowest version available on pip for Python 3.9 and still only work with CUDA compute capability 3.7 (cards such as GeForce TITAN and TITAN BLACK are 3.5). For a list of which GPUs are compatible with which compute capability, see [CUDA GPUs - Compute Capability](https://developer.nvidia.com/cuda-gpus)
+  - To see what compute capability the installed version of torch supports, run: `python -c "import torch; print(torch.cuda.get_arch_list())"`.
 - (Optional) Matlab >= R2016b
 
 

diff --git a/codes/__init__.py b/codes/__init__.py
diff --git a/codes/data/paired_folder_dataset.py b/codes/data/paired_folder_dataset.py
@@ -6,7 +6,7 @@
 import torch
 
 from .base_dataset import BaseDataset
-from utils.base_utils import retrieve_files
+from codes.utils.base_utils import retrieve_files
 
 
 class PairedFolderDataset(BaseDataset):

diff --git a/codes/data/unpaired_folder_dataset.py b/codes/data/unpaired_folder_dataset.py
@@ -6,7 +6,7 @@
 import torch
 
 from .base_dataset import BaseDataset
-from utils.base_utils import retrieve_files
+from codes.utils.base_utils import retrieve_files
 
 
 class UnpairedFolderDataset(BaseDataset):

diff --git a/codes/main.py b/codes/main.py
@@ -5,10 +5,10 @@
 
 import torch
 
-from data import create_dataloader
-from models import define_model
-from metrics import create_metric_calculator
-from utils import dist_utils, base_utils, data_utils
+from codes.data import create_dataloader
+from codes.models import define_model
+from codes.metrics import create_metric_calculator
+from codes.utils import dist_utils, base_utils, data_utils
 
 
 def train(opt):
@@ -224,7 +224,7 @@ def profile(opt, lr_size, test_speed=False):
     msg += f'{"*"*40}\nResolution: {lr_size} -> {hr_size} ({scale}x SR)'
 
     # create model
-    from models.networks import define_generator
+    from codes.models.networks import define_generator
     net_G = define_generator(opt).to(device)
     # base_utils.log_info(f'\n{net_G.__str__()}')
 
@@ -264,7 +264,7 @@ def profile(opt, lr_size, test_speed=False):
     base_utils.log_info(msg)
 
 
-if __name__ == '__main__':
+def main():
     # === parse arguments === #
     args = base_utils.parse_agrs()
 
@@ -290,3 +290,6 @@ def profile(opt, lr_size, test_speed=False):
 
     else:
         raise ValueError(f'Unrecognized mode: {args.mode} (train|test|profile)')
+
+if __name__ == '__main__':
+	main()
diff --git a/codes/metrics/LPIPS/models/__init__.py b/codes/metrics/LPIPS/models/__init__.py
@@ -4,11 +4,14 @@
 from __future__ import print_function
 
 import numpy as np
-from skimage.measure import compare_ssim
+try:
+	from skimage.measure import compare_ssim
+except ImportError:
+	from skimage.metrics import structural_similarity as compare_ssim
 import torch
 from torch.autograd import Variable
 
-from metrics.LPIPS.models import dist_model
+from codes.metrics.LPIPS.models import dist_model
 
 
 class PerceptualLoss(torch.nn.Module):

diff --git a/codes/metrics/LPIPS/models/dist_model.py b/codes/metrics/LPIPS/models/dist_model.py
@@ -19,7 +19,7 @@
 from IPython import embed
 
 from . import networks_basic as networks
-import models as util
+import codes.models as util
 
 class DistModel(BaseModel):
     def name(self):

diff --git a/codes/metrics/LPIPS/models/networks_basic.py b/codes/metrics/LPIPS/models/networks_basic.py
@@ -12,7 +12,7 @@
 from IPython import embed
 from . import pretrained_networks as pn
 
-import metrics.LPIPS.models as util
+import codes.metrics.LPIPS.models as util
 
 def spatial_average(in_tens, keepdim=True):
     return in_tens.mean([2,3],keepdim=keepdim)

diff --git a/codes/metrics/metric_calculator.py b/codes/metrics/metric_calculator.py
@@ -8,8 +8,8 @@
 import torch
 import torch.distributed as dist
 
-from utils import base_utils, data_utils, net_utils
-from utils.dist_utils import master_only
+from codes.utils import base_utils, data_utils, net_utils
+from codes.utils.dist_utils import master_only
 from .LPIPS.models.dist_model import DistModel
 
 

diff --git a/codes/models/base_model.py b/codes/models/base_model.py
@@ -7,8 +7,8 @@
 import torch.distributed as dist
 from torch.nn.parallel import DistributedDataParallel
 
-from utils.data_utils import create_kernel, downsample_bd
-from utils.dist_utils import master_only
+from codes.utils.data_utils import create_kernel, downsample_bd
+from codes.utils.dist_utils import master_only
 
 
 class BaseModel():

diff --git a/codes/models/networks/tecogan_nets.py b/codes/models/networks/tecogan_nets.py
@@ -6,10 +6,10 @@
 import torch.nn.functional as F
 
 from .base_nets import BaseSequenceGenerator, BaseSequenceDiscriminator
-from utils.net_utils import space_to_depth, backward_warp, get_upsampling_func
-from utils.net_utils import initialize_weights
-from utils.data_utils import float32_to_uint8
-from metrics.model_summary import register, parse_model_info
+from codes.utils.net_utils import space_to_depth, backward_warp, get_upsampling_func
+from codes.utils.net_utils import initialize_weights
+from codes.utils.data_utils import float32_to_uint8
+from codes.metrics.model_summary import register, parse_model_info
 
 
 # ====================== generator modules ====================== #
@@ -236,8 +236,8 @@ def step(self, lr_curr, lr_prev, hr_prev):
         lr_flow = self.fnet(lr_curr, lr_prev)
 
         # pad if size is not a multiple of 8
-        pad_h = lr_curr.size(2) - lr_curr.size(2)//8*8
-        pad_w = lr_curr.size(3) - lr_curr.size(3)//8*8
+        pad_h = lr_curr.size(2) - lr_curr.size(2) // 8 * 8
+        pad_w = lr_curr.size(3) - lr_curr.size(3) // 8 * 8
         lr_flow_pad = F.pad(lr_flow, (0, pad_w, 0, pad_h), 'reflect')
 
         # upsample lr flow
@@ -303,8 +303,8 @@ def profile(self, lr_size, device):
         gflops_dict['FNet'], params_dict['FNet'] = parse_model_info(self.fnet)
 
         # profile module 2: sr module
-        pad_h = lr_curr.size(2) - lr_curr.size(2)//8*8
-        pad_w = lr_curr.size(3) - lr_curr.size(3)//8*8
+        pad_h = lr_curr.size(2) - lr_curr.size(2) // 8 * 8
+        pad_w = lr_curr.size(3) - lr_curr.size(3) // 8 * 8
         lr_flow_pad = F.pad(lr_flow, (0, pad_w, 0, pad_h), 'reflect')
         hr_flow = self.scale * self.upsample_func(lr_flow_pad)
         hr_prev_warp = backward_warp(hr_prev, hr_flow)

diff --git a/codes/models/vsr_model.py b/codes/models/vsr_model.py
@@ -6,7 +6,7 @@
 from .base_model import BaseModel
 from .networks import define_generator
 from .optim import define_criterion, define_lr_schedule
-from utils import base_utils, net_utils, data_utils
+from codes.utils import base_utils, net_utils, data_utils
 
 
 class VSRModel(BaseModel):

diff --git a/codes/models/vsrgan_model.py b/codes/models/vsrgan_model.py
@@ -8,7 +8,7 @@
 from .networks import define_generator, define_discriminator
 from .networks.vgg_nets import VGGFeatureExtractor
 from .optim import define_criterion, define_lr_schedule
-from utils import base_utils, net_utils, dist_utils
+from codes.utils import base_utils, net_utils, dist_utils
 
 
 class VSRGANModel(VSRModel):
@@ -243,7 +243,7 @@ def train(self):
         # ping-pong (pp) loss
         if self.pp_crit is not None:
             tempo_extent = self.opt['train']['tempo_extent']
-            hr_data_fw = hr_data[:, :tempo_extent - 1, ...]      # -------->|
+            hr_data_fw = hr_data[:, :tempo_extent - 1, ...]  # -------->|
             hr_data_bw = hr_data[:, tempo_extent:, ...].flip(1)  # <--------|
 
             pp_w = self.opt['train']['pingpong_crit'].get('weight', 1)

diff --git a/codes/official_metrics/LPIPSmodels/util.py b/codes/official_metrics/LPIPSmodels/util.py
@@ -9,7 +9,10 @@
 import collections
 import matplotlib.pyplot as plt
 from scipy.ndimage.interpolation import zoom
-from skimage.measure import compare_ssim
+try:
+	from skimage.measure import compare_ssim
+except ImportError:
+	from skimage.metrics import structural_similarity as compare_ssim
 import torch
 from IPython import embed
 import cv2

diff --git a/codes/official_metrics/evaluate.py b/codes/official_metrics/evaluate.py
@@ -11,7 +11,7 @@
 
     keys = args.model.split('_')
     assert keys[0] in ('TecoGAN', 'FRVSR')
-    assert keys[1] in ('BD', 'BI')
+    assert (keys[1] in ('BD', 'BI')) or (keys[2] in ('BD', 'BI'))
 
     # set dirs
     Vid4_GT_dir = 'data/Vid4/GT'

diff --git a/codes/official_metrics/metrics.py b/codes/official_metrics/metrics.py
@@ -4,7 +4,10 @@
 import pandas as pd
 from LPIPSmodels import util
 import LPIPSmodels.dist_model as dm
-from skimage.measure import compare_ssim
+try:
+	from skimage.measure import compare_ssim
+except ImportError:
+	from skimage.metrics import structural_similarity as compare_ssim
 
 from absl import flags
 flags.DEFINE_string('output', None, 'the path of output directory')