一、Flownet在Ubuntu系統(tǒng)下安裝過程
從[1]中下載代碼
系統(tǒng)環(huán)境
- Ubuntu 18.04.5
cat /etc/issue
- CUDA 10.2.89
cat /usr/local/cuda/version.txt
虛擬環(huán)境
- 先安裝 python 3.7
pip install python==3.7
- 再安裝 torch 1.2.0
pip install torch==1.2.0
- 接著安裝 torchvision 0.4.0
pip install torchvision==0.4.0
- 很重要的一步，安裝 spatial_correlation_sampler-0.2.1
- 先從[2]中下載，并解壓后放入FlowNet文件夾內(nèi)噪猾；flownet文件夾
-
進(jìn)入spatial_correlation_sampler-0.2.1文件夾后，執(zhí)行以下命令：
spatial_correlation_sampler-0.2.1文件夾
python setup.py install
- 其余的安裝包我就列在下面了炉峰，安裝順序無要求
# Name Version Build Channel
_libgcc_mutex 0.1 main
_openmp_mutex 4.5 1_gnu
absl-py 0.13.0 pypi_0 pypi
argparse 1.4.0 pypi_0 pypi
blas 1.0 mkl
bzip2 1.0.8 h7b6447c_0
ca-certificates 2021.7.5 h06a4308_1
cachetools 4.2.2 pypi_0 pypi
cairo 1.14.12 h8948797_3
certifi 2021.5.30 py37h06a4308_0
charset-normalizer 2.0.5 pypi_0 pypi
ffmpeg 4.0 hcdf2ecd_0
fontconfig 2.13.1 h6c09931_0
freeglut 3.0.0 hf484d3e_5
freetype 2.10.4 h5ab3b9f_0
future 0.18.2 pypi_0 pypi
glib 2.63.1 h5a9c865_0
google-auth 1.35.0 pypi_0 pypi
google-auth-oauthlib 0.4.6 pypi_0 pypi
graphite2 1.3.14 h23475e2_0
grpcio 1.40.0 pypi_0 pypi
harfbuzz 1.8.8 hffaf4a1_0
hdf5 1.10.2 hba1933b_1
icu 58.2 he6710b0_3
idna 3.2 pypi_0 pypi
imageio 2.9.0 pypi_0 pypi
importlib-metadata 4.8.1 pypi_0 pypi
intel-openmp 2021.3.0 h06a4308_3350
jasper 2.0.14 h07fcdf6_1
jpeg 9d h7f8727e_0
libedit 3.1.20210714 h7f8727e_0
libffi 3.2.1 hf484d3e_1007
libgcc-ng 9.3.0 h5101ec6_17
libgfortran-ng 7.5.0 ha8ba4b0_17
libgfortran4 7.5.0 ha8ba4b0_17
libglu 9.0.0 hf484d3e_1
libgomp 9.3.0 h5101ec6_17
libopencv 3.4.2 hb342d67_1
libopus 1.3.1 h7b6447c_0
libpng 1.6.37 hbc83047_0
libstdcxx-ng 9.3.0 hd4cf53a_17
libtiff 4.2.0 h85742a9_0
libuuid 1.0.3 h1bed415_2
libvpx 1.7.0 h439df22_0
libwebp-base 1.2.0 h27cfd23_0
libxcb 1.14 h7b6447c_0
libxml2 2.9.12 h03d6c58_0
lz4-c 1.9.3 h295c915_1
markdown 3.3.4 pypi_0 pypi
mkl 2021.3.0 h06a4308_520
mkl-service 2.4.0 py37h7f8727e_0
mkl_fft 1.3.0 py37h42c9631_2
mkl_random 1.2.2 py37h51133e4_0
ncurses 6.2 he6710b0_1
numpy 1.21.2 pypi_0 pypi
numpy-base 1.20.3 py37h74d4b33_0
oauthlib 3.1.1 pypi_0 pypi
opencv 3.4.2 py37h6fd60c2_1
openssl 1.0.2u h7b6447c_0
path 16.2.0 pypi_0 pypi
path-py 12.5.0 pypi_0 pypi
pcre 8.45 h295c915_0
pillow 8.3.2 pypi_0 pypi
pip 21.2.2 py37h06a4308_0
pixman 0.40.0 h7b6447c_0
protobuf 3.18.0 pypi_0 pypi
py-opencv 3.4.2 py37hb342d67_1
pyasn1 0.4.8 pypi_0 pypi
pyasn1-modules 0.2.8 pypi_0 pypi
python 3.7.0 h6e4f718_3
readline 7.0 h7b6447c_5
requests 2.26.0 pypi_0 pypi
requests-oauthlib 1.3.0 pypi_0 pypi
rsa 4.7.2 pypi_0 pypi
scipy 1.7.1 pypi_0 pypi
setuptools 58.0.4 py37h06a4308_0
six 1.16.0 pyhd3eb1b0_0
spatial-correlation-sampler 0.2.1 pypi_0 pypi
sqlite 3.33.0 h62c20be_0
tensorboard 2.6.0 pypi_0 pypi
tensorboard-data-server 0.6.1 pypi_0 pypi
tensorboard-plugin-wit 1.8.0 pypi_0 pypi
tensorboardx 2.4 pypi_0 pypi
tk 8.6.10 hbc83047_0
torch 1.2.0 pypi_0 pypi
torchvision 0.4.0 pypi_0 pypi
tqdm 4.62.2 pypi_0 pypi
typing-extensions 3.10.0.2 pypi_0 pypi
urllib3 1.26.6 pypi_0 pypi
werkzeug 2.0.1 pypi_0 pypi
wheel 0.37.0 pyhd3eb1b0_1
xz 5.2.5 h7b6447c_0
zipp 3.5.0 pypi_0 pypi
zlib 1.2.11 h7b6447c_3
zstd 1.4.9 haebb681_0
- 以上都安裝完成后畏妖，執(zhí)行以下命令，模型就開始訓(xùn)練啦~
python main.py /path/to/flying_chairs/ -b32 -j8 -a flownets
上面的/path/to/flying_chairs/為flying_chairs的路徑
運行5個Epoch時的截圖
二疼阔、Flownet2在Ubuntu系統(tǒng)下安裝過程
從[3]中下載代碼
系統(tǒng)環(huán)境
- Ubuntu 18.04.5
- CUDA 10.2.89
虛擬環(huán)境
- 先安裝 python 3.7
pip install python== 3.7
- 再安裝 torch 1.7.1
pip install torch==1.7.1
- 接著安裝 torchvision 0.8.2
pip install torchvision==0.8.2
- 其余的安裝包我就列在下面了戒劫，安裝順序無要求
# Name Version Build Channel
_libgcc_mutex 0.1 main
_openmp_mutex 4.5 1_gnu
absl-py 0.13.0 pypi_0 pypi
ca-certificates 2021.7.5 h06a4308_1
cachetools 4.2.2 pypi_0 pypi
certifi 2021.5.30 py37h06a4308_0
charset-normalizer 2.0.5 pypi_0 pypi
colorama 0.4.4 pypi_0 pypi
cycler 0.10.0 pypi_0 pypi
google-auth 1.35.0 pypi_0 pypi
google-auth-oauthlib 0.4.6 pypi_0 pypi
grpcio 1.40.0 pypi_0 pypi
idna 3.2 pypi_0 pypi
imageio 2.9.0 pypi_0 pypi
importlib-metadata 4.8.1 pypi_0 pypi
kiwisolver 1.3.2 pypi_0 pypi
ld_impl_linux-64 2.35.1 h7274673_9
libedit 3.1.20210714 h7f8727e_0
libffi 3.2.1 hf484d3e_1007
libgcc-ng 9.3.0 h5101ec6_17
libgomp 9.3.0 h5101ec6_17
libstdcxx-ng 9.3.0 hd4cf53a_17
markdown 3.3.4 pypi_0 pypi
matplotlib 3.4.3 pypi_0 pypi
ncurses 6.2 he6710b0_1
networkx 2.6.3 pypi_0 pypi
numpy 1.21.2 pypi_0 pypi
oauthlib 3.1.1 pypi_0 pypi
openssl 1.0.2u h7b6447c_0
pillow 8.3.2 pypi_0 pypi
pip 21.2.2 py37h06a4308_0
protobuf 3.18.0 pypi_0 pypi
pyasn1 0.4.8 pypi_0 pypi
pyasn1-modules 0.2.8 pypi_0 pypi
pyparsing 2.4.7 pypi_0 pypi
python 3.7.0 h6e4f718_3
python-dateutil 2.8.2 pypi_0 pypi
pytz 2021.1 pypi_0 pypi
pywavelets 1.1.1 pypi_0 pypi
readline 7.0 h7b6447c_5
requests 2.26.0 pypi_0 pypi
requests-oauthlib 1.3.0 pypi_0 pypi
rsa 4.7.2 pypi_0 pypi
scikit-image 0.18.3 pypi_0 pypi
scipy 1.7.1 pypi_0 pypi
setproctitle 1.2.2 pypi_0 pypi
setuptools 58.0.4 py37h06a4308_0
six 1.16.0 pypi_0 pypi
sqlite 3.33.0 h62c20be_0
tensorboard 2.6.0 pypi_0 pypi
tensorboard-data-server 0.6.1 pypi_0 pypi
tensorboard-plugin-wit 1.8.0 pypi_0 pypi
tensorboardx 2.4 pypi_0 pypi
tifffile 2021.8.30 pypi_0 pypi
tk 8.6.10 hbc83047_0
torch 1.7.1 pypi_0 pypi
torchvision 0.8.2 pypi_0 pypi
tqdm 4.62.2 pypi_0 pypi
typing-extensions 3.10.0.2 pypi_0 pypi
tzdata 2021a h5d7bf9c_0
urllib3 1.26.6 pypi_0 pypi
werkzeug 2.0.1 pypi_0 pypi
wheel 0.37.0 pyhd3eb1b0_1
xz 5.2.5 h7b6447c_0
zipp 3.5.0 pypi_0 pypi
zlib 1.2.11 h7b6447c_3
- 很重要的一步，進(jìn)入flownet2文件夾后婆廊，執(zhí)行以下命令：
bash install.sh
- 運行main.py報錯后迅细，修改以下代碼：
from scipy.misc import imread, imresize
為：
from imageio import imread
-
還遇到報錯，就屏蔽掉這幾行代碼：
- 以上都安裝完成后淘邻，執(zhí)行以下命令茵典，模型就開始訓(xùn)練啦~
python main.py --batch_size 8 --model FlowNet2 --loss=L1Loss --optimizer=Adam --optimizer_lr=1e-4 --training_dataset MpiSintelFinal --training_dataset_root /path/to/MPI-Sintel-complete/training/ --validation_dataset MpiSintelClean --validation_dataset_root /path/to/data/MPI-Sintel-complete/training/
上面的/path/to/MPI-Sintel-complete/training/為MPI-Sintel-complete的路徑
參考
[1] https://github.com/ClementPinard/FlowNetPytorch
[2] https://pypi.org/project/spatial-correlation-sampler/0.2.1/
[3] https://github.com/NVIDIA/flownet2-pytorch
