caffe版faster-RCNN环境搭建

faster-rcnn提出论文：《Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks》
faster-rcnn 的算法详解可看这篇博文（清晰易懂，良心博文！）： http://blog.csdn.net/shenxiaolu1984/article/details/51152614
faster-rcnn Python版本源码地址：https://github.com/rbgirshick/py-faster-rcnn

- 基本按照官网上的readme

1.电脑上已经有可运行caffe所需的环境

2.下载faster-rcnn python版本源码

git clone --recursive https://github.com/rbgirshick/py-faster-rcnn.git

3. 进入/py-faster-rcnn/lib 进行编译，build the Cython modules

    cd py-faster-rcnn/lib  
    make

如果重新升级或者降级numpy依赖包，需要重新去掉之前生成的文件，重新make; 否则会报这里面的错

问题：

python setup.py build_ext --inplace
running build_ext
cythoning utils/bbox.pyx to utils/bbox.c

Error compiling Cython file:
------------------------------------------------------------
...
# Copyright (c) 2015 Microsoft
# Licensed under The MIT License [see LICENSE for details]
# Written by Sergey Karayev
# --------------------------------------------------------

cimport cython
       ^
------------------------------------------------------------

utils/bbox.pyx:8:8: Compiler crash in AnalyseDeclarationsTransform

File 'ModuleNode.py', line 122, in analyse_declarations: ModuleNode(bbox.pyx:1:0,
    full_module_name = 'utils.cython_bbox')
File 'Nodes.py', line 408, in analyse_declarations: StatListNode(bbox.pyx:8:0)
File 'Nodes.py', line 408, in analyse_declarations: StatListNode(bbox.pyx:8:8)
File 'Nodes.py', line 7396, in analyse_declarations: CImportStatNode(bbox.pyx:8:8,
    module_name = u'cython')


  File "/home/dsp/anaconda2/lib/python2.7/site-packages/Cython/Utils.py", line 148, in search_include_directories
    path = os.path.join(dir, dotted_filename)
  File "/home/dsp/anaconda2/lib/python2.7/posixpath.py", line 73, in join
    path += '/' + b
UnicodeDecodeError: 'ascii' codec can't decode byte 0xe6 in position 10: ordinal not in range(128)
building 'utils.cython_bbox' extension
gcc -pthread -B /home/dsp/anaconda2/compiler_compat -Wl,--sysroot=/ -fno-strict-aliasing -g -O2 -DNDEBUG -g -fwrapv -O3 -Wall -Wstrict-prototypes -fPIC -I/home/dsp/anaconda2/lib/python2.7/site-packages/numpy/core/include -I/home/dsp/anaconda2/include/python2.7 -c utils/bbox.c -o build/temp.linux-x86_64-2.7/utils/bbox.o -Wno-cpp -Wno-unused-function
utils/bbox.c:1:2: error: #error Do not use this file, it is the result of a failed Cython compilation.
 #error Do not use this file, it is the result of a failed Cython compilation.
  ^
error: command 'gcc' failed with exit status 1
Makefile:2: recipe for target 'all' failed
make: *** [all] Error 1

先各种百度，安装cython等；结果发现时路径的坑
神坑：路径；py-faster-rcnn存放位置路径中不能有中文，否则报错UnicodeDecodeError: 'ascii' codec can't decode byte 0xe6 in position
最后放到 /home/uername/ 下直接makec成功

4.编译/py-faster-rcnn/caffe-fast-rcnn

cd py-faster-rcnn/caffe-fast-rcnn
make -j32 && make pycaffe

Makefile.config文件直接用的本机配置caffe的文件
问题：

dsp@dsp-PowerEdge-R730:/home/user/ran/Desktop/protobuf-2.6.1$ protoc
[libprotobuf FATAL google/protobuf/stubs/common.cc:61] This program requires version 3.4.0 of the Protocol Buffer runtime library, but the installed version is 2.6.1.  Please update your library.  If you compiled the program yourself, make sure that your headers are from the same version of Protocol Buffers as your link-time library.  (Version verification failed in "google/protobuf/any.pb.cc".)
terminate called after throwing an instance of 'google::protobuf::FatalException'
  what():  This program requires version 3.4.0 of the Protocol Buffer runtime library, but the installed version is 2.6.1.  Please update your library.  If you compiled the program yourself, make sure that your headers are from the same version of Protocol Buffers as your link-time library.  (Version verification failed in "google/protobuf/any.pb.cc".)
Aborted (core dumped)

这个问题是服务器上的错误：http://blog.csdn.net/m0_37477175/article/details/78233983

https://www.cnblogs.com/javaee6/p/4849051.html

https://github.com/BVLC/caffe/issues/5711

前面两种方法都是了没有解决，后面github上的issue也没有给出解决方法，看以后能不能解决！！！

In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from src/caffe/blob.cpp:4:
./include/caffe/util/cudnn.hpp:8:34: fatal error: caffe/proto/caffe.pb.h: 没有那个文件或目录

- 按照：解决方法：用protoc从caffe/src/caffe/proto/caffe.proto生成caffe.pb.h和caffe.pb.cc ，http://blog.csdn.net/xmzwlw/article/details/48270225 没有作用；后面还有一串错误

In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from ./include/caffe/layers/loss_layer.hpp:6,
                 from src/caffe/layers/loss_layer.cpp:3:
./include/caffe/util/cudnn.hpp: In function ‘const char* cudnnGetErrorString(cudnnStatus_t)’:
./include/caffe/util/cudnn.hpp:21:10: warning: enumeration value ‘CUDNN_STATUS_RUNTIME_PREREQUISITE_MISSING’ not handled in switch [-Wswitch]
   switch (status) {
          ^
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::setConvolutionDesc(cudnnConvolutionStruct**, cudnnTensorDescriptor_t, cudnnFilterDescriptor_t, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:108:70: error: too few arguments to function ‘cudnnStatus_t cudnnSetConvolution2dDescriptor(cudnnConvolutionDescriptor_t, int, int, int, int, int, int, cudnnConvolutionMode_t, cudnnDataType_t)’
       pad_h, pad_w, stride_h, stride_w, 1, 1, CUDNN_CROSS_CORRELATION));
                                                                      ^
./include/caffe/util/cudnn.hpp:15:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; 
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from ./include/caffe/layers/loss_layer.hpp:6,
                 from src/caffe/layers/loss_layer.cpp:3:
/usr/local/cuda/include/cudnn.h:500:27: note: declared here
 cudnnStatus_t CUDNNWINAPI cudnnSetConvolution2dDescriptor( cudnnConvolutionDescriptor_t convDesc,
                           ^
In file included from ./include/caffe/util/device_alternate.hpp:40:0,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from ./include/caffe/layers/loss_layer.hpp:6,
                 from src/caffe/layers/loss_layer.cpp:3:
./include/caffe/util/cudnn.hpp: In function ‘void caffe::cudnn::createPoolingDesc(cudnnPoolingStruct**, caffe::PoolingParameter_PoolMethod, cudnnPoolingMode_t*, int, int, int, int, int, int)’:
./include/caffe/util/cudnn.hpp:127:41: error: too few arguments to function ‘cudnnStatus_t cudnnSetPooling2dDescriptor(cudnnPoolingDescriptor_t, cudnnPoolingMode_t, cudnnNanPropagation_t, int, int, int, int, int, int)’
         pad_h, pad_w, stride_h, stride_w));
                                         ^
./include/caffe/util/cudnn.hpp:15:28: note: in definition of macro ‘CUDNN_CHECK’
     cudnnStatus_t status = condition; 
                            ^
In file included from ./include/caffe/util/cudnn.hpp:5:0,
                 from ./include/caffe/util/device_alternate.hpp:40,
                 from ./include/caffe/common.hpp:19,
                 from ./include/caffe/blob.hpp:8,
                 from ./include/caffe/layers/loss_layer.hpp:6,
                 from src/caffe/layers/loss_layer.cpp:3:

View Code

后面发现时cndnn版本不兼容：Faster R-CNN 和最新的版本cuDNN V5.0不兼容问题

解决方法：

cd py-faster-rcnn/caffe-fast-rcnn  
Git remote add caffe https://github.com/BVLC/caffe.git  
git fetch caffe  
git merge caffe/master 

在合并之后注释掉include/caffe/layers/python_layer.hppa文件里的self_.attr(“phase”) = static_cast(this->phase_)

重新从caffe中fetch and merge
遇到merge冲突，直接编辑

  // Caffe utility functions
  bp::def("init_log", &InitLog);
  bp::def("init_log", &InitLogLevel);
  bp::def("init_log", &InitLogLevelPipe);
  bp::def("log", &Log);
  bp::def("has_nccl", &HasNCCL);
  bp::def("set_mode_cpu", &set_mode_cpu);
  bp::def("set_mode_gpu", &set_mode_gpu);
  bp::def("set_random_seed", &set_random_seed);
  bp::def("set_device", &Caffe::SetDevice);
<<<<<<< HEAD
  bp::def("set_random_seed", &Caffe::set_random_seed);
=======
  bp::def("solver_count", &Caffe::solver_count);
  bp::def("set_solver_count", &Caffe::set_solver_count);
  bp::def("solver_rank", &Caffe::solver_rank);
  bp::def("set_solver_rank", &Caffe::set_solver_rank);
  bp::def("set_multiprocess", &Caffe::set_multiprocess);
>>>>>>> caffe/master

- 最后编译成功；faster-rcnn 安装里面也提到了处理版本不兼容，进行文件替换的方法

AR -o .build_release/lib/libcaffe.a
LD -o .build_release/lib/libcaffe.so.1.0.0
CXX/LD -o .build_release/tools/convert_imageset.bin
CXX/LD -o .build_release/tools/train_net.bin
CXX/LD -o .build_release/tools/finetune_net.bin
CXX/LD -o .build_release/tools/compute_image_mean.bin
CXX/LD -o .build_release/tools/upgrade_net_proto_text.bin
CXX/LD -o .build_release/tools/test_net.bin
CXX/LD -o .build_release/tools/upgrade_net_proto_binary.bin
CXX/LD -o .build_release/tools/upgrade_solver_proto_text.bin
CXX/LD -o .build_release/tools/net_speed_benchmark.bin
CXX/LD -o .build_release/tools/device_query.bin
CXX/LD -o .build_release/tools/extract_features.bin
CXX/LD -o .build_release/examples/cifar10/convert_cifar_data.bin
CXX/LD -o .build_release/examples/mnist/convert_mnist_data.bin
CXX/LD -o .build_release/examples/cpp_classification/classification.bin
CXX/LD -o .build_release/tools/caffe.bin
CXX/LD -o .build_release/examples/siamese/convert_mnist_siamese_data.bin
CXX/LD -o python/caffe/_caffe.so python/caffe/_caffe.cpp
touch python/caffe/proto/__init__.py
PROTOC (python) src/caffe/proto/caffe.proto

View Code

5、下载Faster R-CNN的预训练模型

cd py-faster-rcnn
./data/scripts/fetch_faster_rcnn_models.sh

model:https://dl.dropboxusercontent.com/s/o6ii098bu51d139/faster_rcnn_models.tgz?dl=0

6、运行demo.py

cd py-faster-rcnn
./tools/demo.py

7.Usage:训练

To train and test a Faster R-CNN detector using the alternating optimization algorithm from our NIPS 2015 paper, use experiments/scripts/faster_rcnn_alt_opt.sh. Output is written underneath $FRCN_ROOT/output.

cd $FRCN_ROOT
./experiments/scripts/faster_rcnn_alt_opt.sh [GPU_ID] [NET] [--set ...]
# GPU_ID is the GPU you want to train on
# NET in {ZF, VGG_CNN_M_1024, VGG16} is the network arch to use
# --set ... allows you to specify fast_rcnn.config options, e.g.
#   --set EXP_DIR seed_rng1701 RNG_SEED 1701

AttributeError: 'module' object has no attribute 'text_format'

在文件./lib/fast_rcnn/train.py增加一行import google.protobuf.text_format 即可解决问题

8.后续问题

I1119 21:29:42.451519 21365 net.cpp:744] Ignoring source layer pool5_spm6
I1119 21:29:42.451537 21365 net.cpp:744] Ignoring source layer pool5_spm6_flatten
I1119 21:29:42.481129 21365 net.cpp:744] Ignoring source layer fc8
I1119 21:29:42.481151 21365 net.cpp:744] Ignoring source layer prob
Solving...
Process Process-3:
Traceback (most recent call last):
  File "/home/dsp/anaconda2/lib/python2.7/multiprocessing/process.py", line 267, in _bootstrap
    self.run()
  File "/home/dsp/anaconda2/lib/python2.7/multiprocessing/process.py", line 114, in run
    self._target(*self._args, **self._kwargs)
  File "./tools/train_faster_rcnn_alt_opt.py", line 195, in train_fast_rcnn
    max_iters=max_iters)
  File "/home/dsp/py-faster-rcnn/tools/../lib/fast_rcnn/train.py", line 162, in train_net
    model_paths = sw.train_model(max_iters)
  File "/home/dsp/py-faster-rcnn/tools/../lib/fast_rcnn/train.py", line 103, in train_model
    self.solver.step(1)
  File "/home/dsp/py-faster-rcnn/tools/../lib/roi_data_layer/layer.py", line 144, in forward
    blobs = self._get_next_minibatch()
  File "/home/dsp/py-faster-rcnn/tools/../lib/roi_data_layer/layer.py", line 63, in _get_next_minibatch
    return get_minibatch(minibatch_db, self._num_classes)
  File "/home/dsp/py-faster-rcnn/tools/../lib/roi_data_layer/minibatch.py", line 55, in get_minibatch
    num_classes)
  File "/home/dsp/py-faster-rcnn/tools/../lib/roi_data_layer/minibatch.py", line 100, in _sample_rois
    fg_inds, size=fg_rois_per_this_image, replace=False)
  File "mtrand.pyx", line 1187, in mtrand.RandomState.choice
TypeError: 'numpy.float64' object cannot be interpreted as an index


TypeError: 'numpy.float64' object cannot be interpreted as an index
还是numpy版本的问题，直接换一个版本好了
sudo pip install -U numpy==1.11.0


--这个错 “ImportError: numpy.core.multiarray failed to import”好像就是lib库里面没有重新make,去掉之前

+ ./tools/train_faster_rcnn_alt_opt.py --gpu 0 --net_name ZF --weights data/imagenet_models/ZF.v2.caffemodel --imdb voc_2007_trainval --cfg experiments/cfgs/faster_rcnn_alt_opt.yml
Traceback (most recent call last):
  File "./tools/train_faster_rcnn_alt_opt.py", line 19, in <module>
    from datasets.factory import get_imdb
  File "/home/dsp/py-faster-rcnn/tools/../lib/datasets/factory.py", line 13, in <module>
    from datasets.coco import coco
  File "/home/dsp/py-faster-rcnn/tools/../lib/datasets/coco.py", line 20, in <module>
    from pycocotools.coco import COCO
  File "/home/dsp/py-faster-rcnn/tools/../lib/pycocotools/coco.py", line 58, in <module>
    import mask
  File "/home/dsp/py-faster-rcnn/tools/../lib/pycocotools/mask.py", line 3, in <module>
    import pycocotools._mask as _mask
  File "pycocotools/_mask.pyx", line 20, in init pycocotools._mask
  File "__init__.pxd", line 989, in numpy.import_array
ImportError: numpy.core.multiarray failed to import

Reading annotation for 4901/4952
Saving cached annotations to /home/dsp/py-faster-rcnn/data/VOCdevkit2007/annotations_cache/annots.pkl
AP for aeroplane = 0.0339
AP for bicycle = 0.1139
AP for bird = 0.0069
AP for boat = 0.0646
Traceback (most recent call last):
  File "./tools/test_net.py", line 90, in <module>
    test_net(net, imdb, max_per_image=args.max_per_image, vis=args.vis)
  File "/home/dsp/py-faster-rcnn/tools/../lib/fast_rcnn/test.py", line 295, in test_net
    imdb.evaluate_detections(all_boxes, output_dir)
  File "/home/dsp/py-faster-rcnn/tools/../lib/datasets/pascal_voc.py", line 322, in evaluate_detections
    self._do_python_eval(output_dir)
  File "/home/dsp/py-faster-rcnn/tools/../lib/datasets/pascal_voc.py", line 285, in _do_python_eval
    use_07_metric=use_07_metric)
  File "/home/dsp/py-faster-rcnn/tools/../lib/datasets/voc_eval.py", line 148, in voc_eval
    BB = BB[sorted_ind, :]
IndexError: too many indices for array

http://blog.csdn.net/weinaonao5418/article/details/54234684

File "/home/dl-box/wei/py-faster-rcnn/tools/../lib/datasets/voc_eval.py", line 149, in voc_eval

BB = BB[sorted_ind, :]

IndexError: too many indices for array

这个错是说没有学习到东西，我用了[4000,2000,4000,2000]试了一下也不行。不过能看到这个错误，如果你的数据集做的没有问题的话，用[80000,40000,80000,40000]（亲测大概16小时，1341张图）运行你自己的数据集就没问题了。为了省时间我用的是[40000,20000,40000,2000]（大概8小时，50张图）。这个时间好像与图片大小和图片数量关系不大。感觉只与迭代次数有关。

Reference

CNN目标检测（一）：Faster RCNN详解

faster-rcnn 安装

Faster R-CNN 的Caffe实现

http://blog.csdn.net/mydear_11000/article/details/70241139

相关阅读:
支持向量机（SVM）
MapReduce(二) MR的高级特性-序列化、排序、分区、合并
 HDFS(二）底层通信原理——RPC 及动态代理
 HDFS（一）高级特性
 UML类图中最重要的几种类关系及其表示
 不同.c文件中声明的全局变量的使用理解
 全局变量和局部变量
 转载永远不要在.h文件中定义变量！
LabVIEW介绍
 C语言程序注释风格
原文地址：https://www.cnblogs.com/ranjiewen/p/7860175.html