/pytorch2caffe

A module to convert pytorch model to caffe model.

Primary LanguagePythonMIT LicenseMIT

pytorch2caffe

A package to convert pytorch model to caffe model.

Note

This package supports F.interpolate with mode='bilinear argument. It depends on caffe with Interp layer. See woodsgao/caffe

Installation

pip3 install pytorch2caffe

or

git clone https://github.com/woodsgao/pytorch2caffe
cd pytorch2caffe
python3 setup.py install

Tutorials

Note: import torch first then import pytorch2caffe

import torch
from torchvision.models import resnet
from pytorch2caffe import pytorch2caffe

name='resnet18'
resnet18=resnet.resnet18()
resnet18.eval()
dummy_input=torch.ones([1,3,224,224])
pytorch2caffe.trans_net(resnet18, dummy_input, name)
pytorch2caffe.save_prototxt('{}.prototxt'.format(name))
pytorch2caffe.save_caffemodel('{}.caffemodel'.format(name))