ubuntu14.04无GPU安装caffe:http://www.cnblogs.com/go-better/p/7160615.html
http://www.jb51.net/article/96169.htm
问题1:编译pycaffe时报错:fatal error: numpy/arrayobject.h没有那个文件或目录
解决:sudo apt-get install python-numpy
(blog.csdn.net/wuzuyu365/article/details/52430657)
问题2:如果编译都通过了(make pycaffe),出现import caffe失败No module named caffe
解决: 把环境变量路径放到 ~/.bashrc文件中, (http://blog.csdn.net/u010417185/article/details/53559107)
打开文件:sudo vim ~/.bashrc
在文件下方写入: export PYTHONPATH=~/caffe/python:$PYTHONPATH
关闭文件使文件生效:source ~/.bashrc
问题3:在使用caffe的python层时经常容易出现如下错误:Check failed: registry.count(type) == 1 (0 vs. 1) Unknown layer type: Python
解决:没有开启对python的支持,需要在Makefile.config文件中开启如下开关:WITH_PYTHON_LAYER=1 然后再make&& make pycaffe
caffe测试mnist手写数字网络:http://blog.csdn.net/lynnandwei/article/details/43273077
测试单张数字:http://blog.csdn.net/xunan003/article/details/73126425
下载mnist图片及二进制文件:http://download.csdn.net/download/liumingchun13/10108641
深度学习目标检测图像数据处理:
图片名批处理:https://www.cnblogs.com/yqyouqing/p/6980243.html
标签工具labellmg: https://github.com/tzutalin/labelImg ( 使用方法:http://blog.csdn.net/jesse_mx/article/details/53606897 )
ImageSets/Main文件夹中的四个txt文件生成:http://blog.csdn.net/gaohuazhao/article/details/60871886
Faster-RCNN测试:
先参考1:http://blog.csdn.net/zyb19931130/article/details/53842791
再参考2:https://www.cnblogs.com/justinzhang/p/5386837.html
YOLOv1论文详解博客:
yolo论文理解(通俗易懂):http://blog.csdn.net/hrsstudy/article/details/70305791 ( hrsstudy博客:http://blog.csdn.net/hrsstudy )
yolo学习笔记(部分模型讲解):http://blog.csdn.net/xjz18298268521/article/details/70037602?locationNum=3&fps=1
yolo测试:http://blog.csdn.net/qq_14845119/article/details/53612362
Yolov2训练及测试:项目主页:https://pjreddie.com/darknet/yolo/
论文翻译:http://lib.csdn.net/article/aimachinelearning/61496
YOLOV2训练自己的数据:http://blog.csdn.net/u010807846/article/details/73554891?locationNum=5&fps=1
测试自己的数据:http://blog.csdn.net/ch_liu23/article/details/53558549
问题:如果测试时出现全屏都是目标框则需要设置测试阈值如
./darknet detect cfg/voc.data cfg/tiny-yolo-voc.cfg backup/final.weights -thresh 0.9
YOLO-VOC预训练模型(权值)下载:http://download.csdn.net/download/liumingchun13/10109794
YOLOv2部分程序解释:http://blog.csdn.net/hysteric314/article/details/54097845
SSD训练及测试:项目主页:https://github.com/weiliu89/caffe/tree/ssd
SSD论文翻译:http://lib.csdn.net/article/deeplearning/53059
SSD论文PPT:http://www.cnblogs.com/lillylin/p/6207292.html(包含论文及代码【Python,C++,caffe】网址, slide【幻灯片】,video)
SSD论文详解:http://blog.csdn.net/u010167269/article/details/52563573
标签工具BBOX-Label-Tool:https://github.com/puzzledqs/BBox-Label-Tool (使用方法:https://www.cnblogs.com/objectDetect/p/5780006.html)
SSD算法caffe配置,训练及测试:http://www.jianshu.com/p/4eaedaeafcb4
SSD安装配置运行:http://lib.csdn.net/article/deeplearning/53859 (详细配置)
http://lib.csdn.net/article/deeplearning/57866 (测试单张图片,百度云盘下载ssd模型)
http://m.blog.csdn.net/majinlei121/article/details/78111023 (detect.py在cpu下运行的修改,路径可以不用改;测试视频将solver_mode = P.Solver.GPU改为solver_mode = P.Solver.CPU)
SSD算法及Caffe代码详解:http://blog.csdn.net/u014380165/article/details/72824889
SSD训练测试自己的数据:https://www.cnblogs.com/EstherLjy/p/6863890.html (data在主目录下创建与ssd中的data不是同一个)
SSD源码解释(Tensorflow):https://zhuanlan.zhihu.com/p/25100992?refer=shanren7
卷积神经网络介绍:
LeNet到DenseNet: https://zhuanlan.zhihu.com/p/31006686
caffe下LeNet详解:http://www.cnblogs.com/denny402/tag/caffe/default.html?page=2(输出参数貌似有问题,与图不对应)