各个版本下载
http://www.vlfeat.org/download/

我是用的VLFeat 0.9.18，但是编译方法通用

1 修改make/nmake_helper.mak

VS2015版本

2 修改Makefile.mak

将Makefile.mak文件中所有出现msvcr的地方改成msvcp（注意，只改小写的地方）

3修改vlfeat-0.9.20/vl/host.h文件

注释掉snprintf isnan

4 命令行编译

最后报这个错其实也是成功的。
NMAKE : fatal error U1077: “echo”: 返回代码“0x1”
Stop.

5 matlab mex 编译

到toolbox目录下，分别运行setup,complie demo

>> vl_setup >> vl_compile vl_compile: assuming that Visual C++ is the active compiler vl_compile: compiling for PCWIN64 (64 bit) MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\aib\vl_aib.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\aib\vl_aibhist.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\fisher\vl_fisher.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\geometry\vl_irodr.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\geometry\vl_rodr.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\gmm\vl_gmm.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\imop\vl_imdisttf.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\imop\vl_imintegral.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\imop\vl_imsmooth.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\imop\vl_imwbackwardmx.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\imop\vl_tpsumx.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\kmeans\vl_hikmeans.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\kmeans\vl_hikmeanspush.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\kmeans\vl_ikmeans.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\kmeans\vl_ikmeanspush.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\kmeans\vl_kmeans.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_alldist.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_alldist2.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_binsearch.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_binsum.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_cummax.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_getpid.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_hog.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_homkermap.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_ihashfind.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_ihashsum.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_inthist.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_kdtreebuild.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_kdtreequery.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_lbp.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_localmax.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_sampleinthist.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_simdctrl.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_svmtrain.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_threads.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_twister.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\misc\vl_version.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\mser\vl_erfill.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\mser\vl_mser.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\quickshift\vl_quickshift.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\sift\vl_covdet.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\sift\vl_dsift.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\sift\vl_liop.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\sift\vl_sift.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\sift\vl_siftdescriptor.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\sift\vl_ubcmatch.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\slic\vl_slic.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 MEX H:\code\dcnf-fcsp-master\libs\vlfeat-0.9.18\toolbox\vlad\vl_vlad.c 使用 'Microsoft Visual C++ 2015 Professional (C)' 编译。 MEX 已成功完成。 >> vl_demo vl_covdet: doubling image: yes vl_covdet: detector: DoG vl_covdet: peak threshold: 0.01, edge threshold: 10 vl_covdet: 13 features suppressed as duplicate (threshold: 0.5) vl_covdet: detected 382 features vl_covdet: kept 360 inside the boundary margin (2) vl_covdet: doubling image: yes vl_covdet: detector: DoG vl_covdet: peak threshold: 0.01, edge threshold: 10 vl_covdet: 13 features suppressed as duplicate (threshold: 0.5) vl_covdet: detected 382 features vl_covdet: kept 360 inside the boundary margin (2) vl_covdet: estimating affine shape for 360 features vl_covdet: 360 features passed affine adaptation vl_covdet: doubling image: yes vl_covdet: detector: DoG vl_covdet: peak threshold: 0.01, edge threshold: 10 vl_covdet: 13 features suppressed as duplicate (threshold: 0.5) vl_covdet: detected 382 features vl_covdet: kept 360 inside the boundary margin (2) vl_covdet: 165 duplicate features were crated due to ambiguous orientation detection (525 total) vl_sift: filter settings: vl_sift: octaves (O) = 5 vl_sift: levels (S) = 3 vl_sift: first octave (o_min) = -1 vl_sift: edge thresh = 10 vl_sift: peak thresh = 0.01 vl_sift: norm thresh = 0 vl_sift: window size = 2 vl_sift: float descriptor = 0 vl_sift: will source frames? no vl_sift: will force orientations? no vl_sift: processing octave -1 vl_sift: processing octave -1 vl_sift: processing octave 0 vl_sift: processing octave 1 vl_sift: processing octave 2 vl_sift: processing octave 3 vl_sift: found 694 keypoints vl_sift: filter settings: vl_sift: octaves (O) = 5 vl_sift: levels (S) = 3 vl_sift: first octave (o_min) = -1 vl_sift: edge thresh = 10 vl_sift: peak thresh = 0 vl_sift: norm thresh = 0 vl_sift: window size = 2 vl_sift: float descriptor = 0 vl_sift: will source frames? yes (638 read) vl_sift: will force orientations? no vl_sift: processing octave -1 vl_sift: processing octave -1 vl_sift: processing octave 0 vl_sift: processing octave 1 vl_sift: processing octave 2 vl_sift: processing octave 3 vl_sift: found 638 keypoints mser: parameters: mser: delta = 1 mser: max_area = 0.75 mser: min_area = 6e-05 mser: max_variation = 0.25 mser: min_diversity = 0.2 mser: statistics: mser: 6 extremal regions of which mser: 5 ( 83.3 % of previous) maximally stable, mser: 5 ( 100 % of previous) stable enough, mser: 5 ( 100 % of previous) small enough, mser: 5 ( 100 % of previous) big enough, mser: 5 ( 100 % of previous) diverse enough. mser: parameters: mser: delta = 32 mser: max_area = 0.75 mser: min_area = 6e-05 mser: max_variation = 0.25 mser: min_diversity = 0.2 mser: statistics: mser: 6 extremal regions of which mser: 5 ( 83.3 % of previous) maximally stable, mser: 4 ( 80 % of previous) stable enough, mser: 4 ( 100 % of previous) small enough, mser: 4 ( 100 % of previous) big enough, mser: 4 ( 100 % of previous) diverse enough. mser: parameters: mser: delta = 159 mser: max_area = 0.75 mser: min_area = 6e-05 mser: max_variation = 0.25 mser: min_diversity = 0.2 mser: statistics: mser: 6 extremal regions of which mser: 5 ( 83.3 % of previous) maximally stable, mser: 1 ( 20 % of previous) stable enough, mser: 1 ( 100 % of previous) small enough, mser: 1 ( 100 % of previous) big enough, mser: 1 ( 100 % of previous) diverse enough. mser: parameters: mser: delta = 160 mser: max_area = 0.75 mser: min_area = 6e-05 mser: max_variation = 0.25 mser: min_diversity = 0.2 mser: statistics: mser: 6 extremal regions of which mser: 5 ( 83.3 % of previous) maximally stable, mser: 0 ( 0 % of previous) stable enough, mser: 0 ( 0 % of previous) small enough, mser: 0 ( 0 % of previous) big enough, mser: 0 ( 0 % of previous) diverse enough. 1/10 113096 regions ...

成功

ref
https://blog.csdn.net/weiwei9363/article/details/65434976
https://blog.csdn.net/jizhidexiaolili/article/details/79810342

总结

以上是生活随笔为你收集整理的VLfeat win10 vs2015 matlab编译的全部内容，希望文章能够帮你解决所遇到的问题。

如果觉得生活随笔网站内容还不错，欢迎将生活随笔推荐给好友。