说明:这个是Faster RCNN刚出来时候的博文记录,最新的可能会有更变,如有问题,请大家查阅官网链接。
先上个检测效果:
(1)图片人脸检测+关键点
(2)摄像头实时人脸+关键点
*************************************************************************
安装
************************************************************************
###1
解压:
py-faster-rcnn-master.zip下载 解压到 py-faster-rcnn;
caffe-faster-rcnn.zip下载 解压到 caffe-faster-rcnn
替换:
用解压的 caffe-faster-rcnn 替换 py-faster-rcnn/caffe-faster-rcnn
###2
修改 py-faster-rcnn/caffe-faster-rcnn/Makefile.config下载参考
# USE_CUDNN := 1 (我默认是关闭的)
MATLAB_DIR、PYTHON_INCLUDE、cuda计算能力和路径
WITH_PYTHON_LAYER := 1
###3
检查安装依赖项
pip install cython
sudo apt-get install python-opencv
pip install easydict
###4
编译Cython modules
cd py-faster-rcnn/lib
make
###5
编译 Caffe and pycaffe
cd py-faster-rcnn/caffe-fast-rcn
make -j8 && make pycaffe
###6
下载预训练模型,解压到 py-faster-rcnn/data
cd py-faster-rcnn/
./data/scripts/fetch_faster_rcnn_models.sh
This will populate the `py-faster-rcnn/data` folder with `faster_rcnn_models`.
These models were trained on VOC 2007 trainval.
*************************************************************************
训练
*************************************************************************
###1
制作数据集目录格式
删除:
(1)data/VOCdevkit2007/VOC2007下所有文件
新建
在 ./data/VOCdevkit2007/VOC2007新建 Annotations;ImageSets/Main;JPEGImages
说明:
Annotations: 保存标签txt转换的xml文件
JPEGImages: 图片文件
ImageSets/Main:文件名列表(不含后缀)
训练集: train.txt
训练验证集: trainval.txt
测试集: test.txt
验证集: val.txt
#Annotations举例:
data/VOCdevkit2007/VOC2007/Annotations/0_1_5.xml
内容格式:
<annotation>
<folder>VOC2007</folder>
<filename>0_1_5.jpg</filename>
<source>
<database>My Database</database>
<annotation>VOC2007</annotation>
<image>flickr</image>
<flickrid>NULL</flickrid>
</source>
<owner>
<flickrid>NULL</flickrid>
<name>deeplearning</name>
</owner>
<size>
<width>160</width>
<height>216</height>
<depth>3</depth>
</size>
<segmented>0</segmented>
<object>
<name>1</name>
<pose>Unspecified</pose>
<truncated>0</truncated>
<difficult>0</difficult>
<bndbox>
<xmin>48</xmin>
<ymin>48</ymin>
<xmax>107</xmax>
<ymax>107</ymax>
</bndbox>
</object>
</annotation>
###2
修改接口
#(1) 修改prototxt配置文件
models/pascal_voc/ZF/faster_rcnn_alt_opt文件夹下的5个文件,分别为
stage1_rpn_train.pt、stage1_fast_rcnn_train.pt、
stage2_rpn_train.pt、stage2_fast_rcnn_train.pt和fast_rcnn_test.pt
① stage1_fast_rcnn_train.pt、stage2_fast_rcnn_train.pt
修改3个参数
num_class:2(识别1类+背景1类)
cls_score中num_output:2
bbox_pred中num_output:8
② stage1_rpn_train.pt、stage2_rpn_train.pt
修改1个参数
num_class:2(识别1类+背景1类)
③ fast_rcnn_test.pt
修改2个参数:
cls_score中num_output:2
bbox_pred中num_output:8
#(2) 修改lib/datasets/pascal_voc.py
self._classes = ('__background__', # always index 0
'people')(只有这一类)
#(3) 修改lib/datasets/imdb.py
该文件的append_flipped_images(self)函数
widths = [PIL.Image.open(self.image_path_at(i)).size[0]
for i in xrange(num_images)]
在 boxes[:, 2] = widths[i] - oldx1 - 1下加入代码:
for b in range(len(boxes)):
if boxes[b][2]< boxes[b][0]:
boxes[b][0] = 0
#(4) 修改完pascal_voc.py和imdb.py后进入lib/datasets目录下删除原来的pascal_voc.pyc和imdb.pyc文件,重新生成这两个文件,因为这两个文件是python编译后的文件,系统会直接调用。
终端进入lib/datasets文件目录输入:
python(此处应出现python的版本)
>>>importpy_compile
>>>py_compile.compile(r'imdb.py')
>>>py_compile.compile(r'pascal_voc.py')
#(5) 删除缓存文件
① 删除output/
② 删除py-faster-rcnn/data/cache中的文件和
py-faster-rcnn/data/VOCdevkit2007/annotations_cache中的文件删除。
#(6) 调参
① 学习率等之类的设置:
py-faster-rcnn/models/pascal_voc/ZF/faster_rcnn_alt_opt中的solve文件设置
② 迭代次数:
py-faster-rcnn/tools/train_faster_rcnn_alt_opt.py中修改
py-faster-rcnn/models/pascal_voc/ZF/faster_rcnn_alt_opt里对应的solver文件(有4个)也修改,stepsize小于上面修改的数值。
#(7) 训练
./experiments/scripts/faster_rcnn_alt_opt.sh 0 ZF pascal_voc
*************************************************************************
测试
*************************************************************************
#(1) 训练完成之后,将output/faster_rcnn_alt_opt/voc_2007_trainval中的最终模型ZF_faster_rcnn_final.caffemodel拷贝到data/faster_rcnn_models中。
#(2) 修改/tools/demo.py:
① CLASSES =('__background__',
'people')
② NETS ={'vgg16': ('VGG16',
'VGG16_faster_rcnn_final.caffemodel'),
'zf': ('ZF',
'ZF_faster_rcnn_final.caffemodel')}
#(3) 在训练集图片中找一张出来放入py-faster-rcnn/data/demo文件夹中,命名为000001.jpg。
im_names = ['000001.jpg']
for im_name in im_names:
print '~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~'
print 'Demo for data/demo/{}'.format(im_name)
demo(net, im_name)
#(4) 运行demo,即在py-faster-rcnn文件夹下终端输入:
./tools/demo.py --net zf</span>
#(5) 或者将默认的模型改为zf:
parser.add_argument('--net', dest='demo_net', help='Network to use [vgg16]',
choices=NETS.keys(), default='vgg16')
修改:
default='zf'
执行:
./tools/demo.py
*************************************************************************
错误调试
*************************************************************************
error 1:assert (boxes[:, 2] >= boxes[:, 0]).all()
Process Process-1:
Traceback (most recent call last):
File "/usr/lib/python2.7/multiprocessing/process.py", line 258, in _bootstrap
self.run()
File "/usr/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 123, in train_rpn
roidb, imdb = get_roidb(imdb_name)
File "./tools/train_faster_rcnn_alt_opt.py", line 68, in get_roidb
roidb = get_training_roidb(imdb)
File "/home/microway/test/pytest/py-faster-rcnn/tools/../lib/fast_rcnn/train.py", line 121, in get_training_roidb
imdb.append_flipped_images()
File "/home/microway/test/pytest/py-faster-rcnn/tools/../lib/datasets/imdb.py", line 108, in append_flipped_images
assert (boxes[:, 2] >= boxes[:, 0]).all()
AssertionError
error1 解决办法:
将py-faster-rcnn/lib/datasets/imdb.py中的相应代码改成如下代码即可:
def append_flipped_images(self):
num_images = self.num_images
widths = [PIL.Image.open(self.image_path_at(i)).size[0]
for i in xrange(num_images)]
for i in xrange(num_images):
boxes = self.roidb[i]['boxes'].copy()
oldx1 = boxes[:, 0].copy()
oldx2 = boxes[:, 2].copy()
boxes[:, 0] = widths[i] - oldx2 - 1
boxes[:, 2] = widths[i] - oldx1 - 1
for b in range(len(boxes)):
if boxes[b][2] < boxes[b][0]:
boxes[b][0] = 0
assert (boxes[:, 2] >= boxes[:, 0]).all()
error 2:IndexError: list index out of range
File "./tools/train_net.py", line 85, in
roidb = get_training_roidb(imdb)
File "/usr/local/fast-rcnn/tools/../lib/fast_rcnn/train.py", line 111, in get_training_roidb
rdl_roidb.prepare_roidb(imdb)
File "/usr/local/fast-rcnn/tools/../lib/roi_data_layer/roidb.py", line 23, in prepare_roidb
roidb[i]['image'] = imdb.image_path_at(i)
IndexError: list index out of range
error2 解决办法:
删除fast-rcnn-master/data/cache/ 文件夹下的.pkl文件,或者改名备份,重新训练即可。
参考资料:
https://github.com/rbgirshick/py-faster-rcnn/issues/34
https://github.com/rbgirshick/fast-rcnn/issues/79