关于在安装caffe2环境中遇到的坑整理（欢迎入坑讨论）

---

1.ImportError: cannot import name caffe2_pb2

测试caffe2的pytorch环境是否正常的时候使用

root@lxsj-ThinkStation:~/pytorch# python
Python 2.7.12 (default, Dec  4 2017, 14:50:18) 
[GCC 5.4.0 20160609] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> from caffe2.python import workspace
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "caffe2/python/__init__.py", line 2, in <module>
    from caffe2.proto import caffe2_pb2
ImportError: cannot import name caffe2_pb2

解决方案：在pytorch路径下，或者root目录下找到.bashrc文件，打开。

1. 在末尾fi后面 写入

export PYTHONPATH=/root/pytorch/build:/usr/local:$PYTHONPATH
export LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH

 然后

source .bashrc

重新进入之后不会报错了

 

---

2.ImportError: No module named _tkinter, please install the python-tk package

安装python-tk即可

apt-get install python-tk

 

---

3.WARNING:root:This caffe2 python run does not have GPU support. Will run in CPU only mode

root@lxsj-ThinkStation:~/download#  python -c 'from caffe2.python import core'
WARNING:root:This caffe2 python run does not have GPU support. Will run in CPU only mode.

然后需要在环境变量里添加一行 vim ~/.bashrc

export PATH=/usr/local/cuda/bin${PATH:+:${PATH}}

然后source ~/.bashrc

---

4.AssertionError: Torch not compiled with CUDA enabled

执行如下脚本

import os
import numpy as np
 
import torch
import torch.nn
import torchvision.models as models
from torch.autograd import Variable 
import torch.cuda
import torchvision.transforms as transforms
 
from PIL import Image
 
img_to_tensor = transforms.ToTensor()
 
def make_model():
    resmodel=models.resnet34(pretrained=True)
    resmodel.cuda()#将模型从CPU发送到GPU,如果没有GPU则删除该行
    return resmodel
 
#分类
def inference(resmodel,imgpath):
    resmodel.eval()#必需，否则预测结果是错误的
    
    img=Image.open(imgpath)
    img=img.resize((224,224))
    tensor=img_to_tensor(img)
    
    tensor=tensor.resize_(1,3,224,224)
    tensor=tensor.cuda()#将数据发送到GPU，数据和模型在同一个设备上运行
            
    result=resmodel(Variable(tensor))
    result_npy=result.data.cpu().numpy()#将结果传到CPU，并转换为numpy格式
    max_index=np.argmax(result_npy[0])
    
    return max_index
    
#特征提取
def extract_feature(resmodel,imgpath):
    resmodel.fc=torch.nn.LeakyReLU(0.1)
    resmodel.eval()
    
    img=Image.open(imgpath)
    img=img.resize((224,224))
    tensor=img_to_tensor(img)
    
    tensor=tensor.resize_(1,3,224,224)
    tensor=tensor.cuda()
            
    result=resmodel(Variable(tensor))
    result_npy=result.data.cpu().numpy()
    
    return result_npy[0]
    
if __name__=="__main__":
    model=make_model()
    imgpath='xx.jpg'
    print inference(model,imgpath)
    print extract_feature(model, imgpath)

结果提示

root@lxsj-ThinkStation:~/example# python index.py 
Traceback (most recent call last):
  File "index.py", line 57, in <module>
    model=make_model()
  File "index.py", line 19, in make_model
    resmodel.cuda()#将模型从CPU发送到GPU,如果没有GPU则删除该行
  File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/module.py", line 258, in cuda
    return self._apply(lambda t: t.cuda(device))
  File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/module.py", line 185, in _apply
    module._apply(fn)
  File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/module.py", line 191, in _apply
    param.data = fn(param.data)
  File "/usr/local/lib/python2.7/dist-packages/torch/nn/modules/module.py", line 258, in <lambda>
    return self._apply(lambda t: t.cuda(device))
  File "/usr/local/lib/python2.7/dist-packages/torch/cuda/__init__.py", line 161, in _lazy_init
    _check_driver()
  File "/usr/local/lib/python2.7/dist-packages/torch/cuda/__init__.py", line 75, in _check_driver
    raise AssertionError("Torch not compiled with CUDA enabled")
AssertionError: Torch not compiled with CUDA enabled

那么我们就来检测cuda是否生效

root@lxsj-ThinkStation:~/example# cat test.py 
import torch
print(torch.cuda.is_available())

返回的是False

所以估计是当时编译的pytorch的方式有问题，所以重新编译

cd pytorch
git submodule update --init --recursive
python setup.py install

然后运行测试cuda返回True，运行刚才的脚本不再抛出异常

5.系统找不到caffe2的依赖库和头文件

应该是当时只编译了，要进入到build目录下进行make install 然后重新编译即可