Make it work on windows
-
ensure you have installed required components
- vs 2017 / vs 2019
- cuda
- python 38 and all required python dependencies
-
open
x64 Native Tools Command Prompt for VS 2019command prompt -
navigate to the root folder of
warp-ctc, simply runbuild-win.cmd
-
if everything is OK, you'll get the the
warpctc.dllin thebuildfolder
Assume you're using conda or miniconda
-
activate python env
conda activate py38
-
navigate to the
warp-ctc\pytorch_bindingfolder -
run command
python setup.py install
-
if everything is OK, copy
warpctc.dllto your python env's root folder, for example:Miniconda3\envs\py38, so python can find and load it
-
navigate to somewhere except the
warpctc\pytorch_bindingfolder, otherwise you'll get errors like_warp_ctc module not found -
start a python cli and inputs
import warpctc_pytorch
My ENV for your reference:
- windows 10 (19041.572)
- vs 2019
- miniconda (py38)
- cuda 11.0
Note: the modification is based on https://github.com/hzli-ucas/warp-ctc#specific-modifications
This is an extension onto the original repo found here.
Install PyTorch v0.4.
WARP_CTC_PATH should be set to the location of a built WarpCTC
(i.e. libwarpctc.so). This defaults to ../build, so from within a
new warp-ctc clone you could build WarpCTC like this:
git clone https://github.com/SeanNaren/warp-ctc.git
cd warp-ctc
mkdir build; cd build
cmake ..
makeNow install the bindings:
cd pytorch_binding
python setup.py installIf you try the above and get a dlopen error on OSX with anaconda3 (as recommended by pytorch):
cd ../pytorch_binding
python setup.py install
cd ../build
cp libwarpctc.dylib /Users/$WHOAMI/anaconda3/libThis will resolve the library not loaded error. This can be easily modified to work with other python installs if needed.
Example to use the bindings below.
import torch
from warpctc_pytorch import CTCLoss
ctc_loss = CTCLoss()
# expected shape of seqLength x batchSize x alphabet_size
probs = torch.FloatTensor([[[0.1, 0.6, 0.1, 0.1, 0.1], [0.1, 0.1, 0.6, 0.1, 0.1]]]).transpose(0, 1).contiguous()
labels = torch.IntTensor([1, 2])
label_sizes = torch.IntTensor([2])
probs_sizes = torch.IntTensor([2])
probs.requires_grad_(True) # tells autograd to compute gradients for probs
cost = ctc_loss(probs, labels, probs_sizes, label_sizes)
cost.backward()CTCLoss(size_average=False, length_average=False)
# size_average (bool): normalize the loss by the batch size (default: False)
# length_average (bool): normalize the loss by the total number of frames in the batch. If True, supersedes size_average (default: False)
forward(acts, labels, act_lens, label_lens)
# acts: Tensor of (seqLength x batch x outputDim) containing output activations from network (before softmax)
# labels: 1 dimensional Tensor containing all the targets of the batch in one large sequence
# act_lens: Tensor of size (batch) containing size of each output sequence from the network
# label_lens: Tensor of (batch) containing label length of each example