Talk Is Cheap

可理解为什么最开始的一层data和label都是top等等 https://github.com/BVLC/caffe/blob/master/examples/mnist/lenet_train_test.prototxt

http://ethereon.github.io/netscope/quickstart.html

运行caffe的faster rcnn的./tools/demo.py 只报了一行的错 Segmentation fault (core dumped) 用print来debug 最后发现是import cv2的错 这样就有一些思路了

报错内容：LD -o .build_release/lib/libcaffe.so.1.0.0-rc3/usr/bin/ld: cannot find -lcblas/usr/bin/ld: cannot find -latlascollect2: error: ld returned 1 exit statusmake: *** [.build_release/lib/libcaffe.s

先运行个yum -y install tcl-devel tk-devel，不知道有没有用 下载了tk和tcl的.tar.gz文件 这样装上它们 cd unix ./configure --prefix="/home/123/123" make make install 然后要重装python 这时要编辑python的Modules/Setup.dist文件的tk相关的地方，

https://github.com/BVLC/caffe/issues/1799

/home/gt/py-faster-rcnn/caffe-fast-rcnn/build/tools/caffe: error while loading shared libraries: libopencv_core.so.3.0: cannot open shared object file: No such file or directory/usr/local/lib里有libopenc

在一台linux上启动caffe mnist lenet程序，同时nsight调试进入cuda代码里