![](https://img-blog.csdnimg.cn/20201014180756918.png?x-oss-process=image/resize,m_fixed,h_64,w_64)
caffe
Xinshuo Webg
一入vision深似海
展开
-
Tips for preparing input data in caffe
For ImageDataLayer, image_data_param will only perform on data. The label won't be affected and should be the second top layer in the layer definition.原创 2017-01-28 12:23:15 · 311 阅读 · 0 评论 -
Training parameter explanation in caffe
RepostBatch Size Batch size in mainly depended to your memory in GPU/RAM. Most time it is used power of two (64,128,256). I always try to choose 256, because it works better with SGD. But for bigger原创 2017-01-28 12:36:39 · 273 阅读 · 0 评论 -
caffe self-customized log displaying method
add LOG(INFO) << “”; to the caffe source code原创 2017-01-31 08:44:29 · 373 阅读 · 0 评论 -
label for caffe image data layer
caffe image data layer can only load integer label due to the function it uses (atoi) in the source code. And also one single label. So it might not be suitable for regression problem.原创 2017-02-01 02:50:59 · 762 阅读 · 0 评论 -
caffe compilation: libopencv_highgui undefined reference to TIFF library
The problem is caused by adding some include and library to makefile.config.One should eliminate the one which actually cause the problem and delete it. And reboot is required to solve this.原创 2017-03-08 02:32:47 · 2050 阅读 · 2 评论 -
matcaffe compilation problem: undefined reference to mxIsGPUArray
Even if the mxGPUArray.h is added to the include path. The library is not linked. Solution is following:https://www.mathworks.com/matlabcentral/answers/294938-cannot-find-lmwgpuThe error message is as原创 2017-03-08 10:07:06 · 1559 阅读 · 5 评论 -
pycaffe problem: loss doesn't display correctly by using python loss layer on caffe
In general, the caffe log will print out two losses: 1. Iteration x, loss = … 2. Train net output #x: loss = …The problem is only the second loss displays correctly when we use a python loss原创 2017-03-21 07:58:29 · 517 阅读 · 0 评论