Caffe
vwenyu-L
这个作者很懒,什么都没留下…
展开
-
caffe --- mistakes
1. Check failed: img_height == data_mean_.height() (224 vs. 256)layer { name: "data" type: "ImageData" top: "data" top: "label" include { phase: TRAIN } transform_param {原创 2016-03-30 21:03:07 · 883 阅读 · 2 评论 -
py-faster-rcnn
--------------------reference-----------------------0. https://github.com/rbgirshick/py-faster-rcnn1. http://blog.csdn.net/samylee/article/details/512017442. http://blog.csdn.net/hgd1010101/arti原创 2016-10-02 11:11:30 · 519 阅读 · 0 评论 -
c++-file about
#includeifstream infile(filename.c_str());while( getline( infile, line) ) { }while( infile >> first_position >> second_pos >> ... ) { }原创 2016-10-03 11:28:58 · 553 阅读 · 0 评论 -
caffe-source code
http://www.zhihu.com/question/27982282原创 2016-04-20 09:59:01 · 409 阅读 · 0 评论 -
caffe-custom network input/image_data_layer/hdf5_data_layer
当需要输入定制自己的数据时候,e.g. multi-label, bbox. etc. option 1. 修改 image_data_layer.cpp 文件 batch类有data_ label_ 两个成员变量, 把图片之外的数据输入到label中就行了test.jpg 1 2 3 4 5vector> > lines原创 2016-10-03 16:26:58 · 718 阅读 · 0 评论 -
caffe-slice layer
usage:layer { name: "slicex" type: "Slice" bottom: "rnn1" top: "x1" top: "x2" top: "x3" slice_param { axis: 0 #slice_point: x #if no this slice_point, it will slice the data depend on n原创 2016-10-05 21:46:20 · 1488 阅读 · 0 评论 -
caffe-scale layer
caffe-scale layer原创 2016-10-05 20:01:11 · 1109 阅读 · 0 评论 -
caffe-reshape layer
reshape layer原创 2016-10-05 18:34:20 · 3747 阅读 · 0 评论 -
caffe-innerproduct layer
caffe-innerproduct原创 2016-10-06 10:44:55 · 707 阅读 · 0 评论 -
google protobuffer
https://developers.google.com/protocol-buffers/docs/cpptutorial#why-use-protocol-buffershttps://developers.google.com/protocol-buffers/docs/pythontutorial#reading-a-messageProtocol B原创 2016-04-20 13:59:17 · 1011 阅读 · 0 评论 -
sublime-shortcuts
0. shift + right-click.原创 2016-10-02 16:54:56 · 223 阅读 · 0 评论 -
caffe-python apis
1. caffe.['AdaDeltaSolver', 'AdaGradSolver', 'AdamSolver', 'Classifier', 'Detector', 'Layer', 'NesterovSolver', 'Net', 'NetSpec', 'RMSPropSolver', 'SGDSolver', 'TEST', 'TRAIN', '_caffe'原创 2017-04-28 15:59:44 · 500 阅读 · 0 评论 -
flownet2-errors
1. engine error'Layer conv1 has unknown engine'modify: CUDNN -> CAFFEhttps://github.com/BVLC/caffe/issues/26822. ‘RandomGeneratorParameter’ does not name a type, ’https://github.com/l原创 2017-07-28 15:24:35 · 1314 阅读 · 0 评论 -
caffe-python layer
usage: layer { name: "data" type: "Python" top: "data" top: "label" top: "cont" python_param { module: "pythondatalayer" layer: "gene_data" param_str: " {'batch_size':10, 'sh原创 2016-10-23 16:14:10 · 1990 阅读 · 0 评论 -
usage-of-caffe
1. training stage ----- log -----xxx.sh!/bin/bashcaff_root=xxxsolver_file = xxxcaffe_file = caffe_root+'built/tools/caffe$caffe train --solver $solver_file | tee $log_file----------------原创 2016-03-10 09:19:16 · 899 阅读 · 0 评论 -
caffe+ubuntu14.04 (without GPU)
http://caffe.berkeleyvision.org/https://github.com/BVLC/caffe0.-sudo apt-get install build-essential-sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libbo原创 2016-02-27 19:20:41 · 556 阅读 · 0 评论 -
caffe-models/resource
http://www.robots.ox.ac.uk/~vgg/research/very_deep/https://github.com/BVLC/caffe/wiki/Model-Zoo#models-used-by-the-vgg-team-in-ilsvrc-2014原创 2016-06-08 11:21:56 · 332 阅读 · 0 评论 -
python-lmdb
-----------write---------------import lmdbenv_db = lmdb.Environment('path-to-save', xxx)txn = env_db.begin( write=True, buffer=True)txn.put( key, value)txn.commit()env_db.close()----原创 2016-04-27 20:58:49 · 1831 阅读 · 0 评论 -
C++-glog/gflags/protobuf
C++-glog/gflags原创 2016-07-24 14:23:52 · 2399 阅读 · 0 评论 -
caffe源码-conv_layer
先来一下code, 基本思路就是把 input_shape( c x h x w ) -> (out_h x out_w) x (c x k_h x k_w), 把filter(o_c x i_c x k_h x k_w)-> o_c x (i_c x k_h x k_w) -------------reference---------------原创 2016-07-20 20:32:06 · 283 阅读 · 0 评论 -
wincaffe-error
nuget神器, 多看看CommonSettings.props文件, 以及vs2013工程管理的方式。。(哭。。。)Restoring NuGet packages...To prevent NuGet from restoring packages during build, open the Visual Studio Options dialog, click on the原创 2016-10-09 10:15:07 · 2608 阅读 · 0 评论 -
caffe-reduction layer
Reduction原创 2016-10-05 20:07:59 · 3250 阅读 · 0 评论 -
caffe-siamese network
有参数的layerparam { name: "xxx"}// Specifies training parameters (multipliers on global learning constants,// and the name and other settings used for weight sharing).mess原创 2016-10-09 20:24:41 · 1238 阅读 · 0 评论 -
caffe-recurrent/rnn
下面的prototxt展示了 RNN的基本用法bottom[0] t * n .... databottom[1] t * n [0, 1] indicator- whether continuous or not. [1, 0]bottom[2] constant input to all cellthe way unroll the rnn, refer to / read log and codeparameter[0 ...4] : W_xh b_h W_hh W原创 2016-10-04 17:22:59 · 1205 阅读 · 2 评论 -
caffe-net construct flow
1. when init a net instance will call a Init functemplate Net::Net(const NetParameter& param, const Net* root_net) : root_net_(root_net) { Init(param);}2. init funci. call layers Setup f原创 2016-10-18 13:55:22 · 345 阅读 · 0 评论 -
caffe-custom new layer
基本就是添加new_layer.cpp, new_layer.hpp, new_layer.cu三个文件:1. new_layer.hppNetLayer()构造函数LayerSetup()函数Reshape()函数type()层名函数Forward_cpu()Forward_gpu()Backward_cpu()Backward_gpu()原创 2016-07-24 12:43:39 · 772 阅读 · 0 评论 -
caffe-dummy data layer
generate data randomlylayer { type: 'DummyData' name: 'x' top: 'x' dummy_data_param { shape: { dim: 10 dim: 3 dim: 2 } data_filler: { type: 'gaussian' } }}原创 2016-10-23 15:54:29 · 4046 阅读 · 0 评论 -
caffe-windows
--------------------------reference-------------------------------https://github.com/BVLC/caffe/tree/windows原创 2016-09-30 18:09:22 · 254 阅读 · 0 评论