hyseob / mdnet

Learning Multi-Domain Convolutional Neural Networks for Visual Tracking

MDNet: Multi-Domain Convolutional Neural Network Tracker

Created by Hyeonseob Nam and Bohyung Han at POSTECH

Project Webpage: http://cvlab.postech.ac.kr/research/mdnet/

News

(May 28, 2017) Python implementation of MDNet is avaliable! [py-MDNet]

Introduction

MDNet is the state-of-the-art visual tracker based on a CNN trained on a large set of tracking sequences, and the winner tracker of The VOT2015 Challenge.

Detailed description of the system is provided by our paper.

This software is implemented using MatConvNet and part of R-CNN.

Citation

If you're using this code in a publication, please cite our paper.

@InProceedings{nam2016mdnet,
author = {Nam, Hyeonseob and Han, Bohyung},
title = {Learning Multi-Domain Convolutional Neural Networks for Visual Tracking},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2016}
}

License

This software is being made available for research purpose only. Check LICENSE file for details.

System Requirements

This code is tested on 64 bit Linux (Ubuntu 14.04 LTS).

Prerequisites 0. MATLAB (tested with R2014a) 0. MatConvNet (tested with version 1.0-beta10, included in this repository) 0. For GPU support, a GPU (~2GB memory) and CUDA toolkit according to the MatConvNet installation guideline will be needed.

Installation

0. Compile MatConvNet according to the installation guideline. An example script is provided in 'compile_matconvnet.m'.
1. Run 'setup_mdnet.m' to set the environment for running MDNet.

Online Tracking using MDNet

Pretrained Models

If you only need to run the tracker, you can use the pretrained MDNet models: 0. models/mdnet_vot-otb.mat (trained on VOT13,14,15 excluding OTB) 0. models/mdnet_otb-vot14.mat (trained on OTB excluding VOT14) 0. models/mdnet_otb-vot15.mat (trained on OTB excluding VOT15)

Demo 0. Run 'tracking/demo_tracking.m'.

The demo performs online tracking on 'Diving' sequence using a pretrained model 'models/mdnet_vot-otb.mat'.

In case of out of GPU memory, decrease opts.batchSize_test in 'tracking/mdnet_init.m'. You can also disable the GPU support by setting opts.useGpu in 'tracking/mdnet_init.m' to false (not recommended).

Learning MDNet

Preparing Datasets

You may need OTB and VOT datasets for learning MDNet models. You can also use other datasets by configuring 'utils/genConfig.m'. 0. Download OTB and VOT datasets. 0. Locate the OTB sequences in 'dataset/OTB' and VOT201x sequences in 'dataset/VOT/201x', or modify the variables benchmarkSeqHome in 'utils/genConfig.m' properly.

Demo 0. Run 'pretraining/demo_pretraining.m'.

The demo trains new MDNet models using OTB or VOT sequences.

mdnet's Issues

question about the input size

@HyeonseobNam
Thanks for your generous to make your excellent work open.
Recently, when i read your paper, i am confused with your network architecture. In your paper, you said that the input size is "107 = 75 (receptive) + 216 (stride)". Can you explain me how to get the "75" and "216"?
Thank you very much again.

How to test on a random video sequence?

Given one video sequence(not in VOT or OTB) and the bounding box of target in first frame, how to write code for tracking task?
Thanks a lot.

vl_nnconv

vl_nnconv
nnconv_forward_blas<>: : gemm<>: [cublas error] [cublas:cublasSgemm (CUBLAS_STATUS_EXECUTION_FAILED)]

How to run the demo

@HyeonseobNam
Hi, I have compiled the MatConvNet and run 'setup_mdnet.m' to set the environment for running MDNet, but when I try to run the demo it didn't work, I felt confused.
The MatConvNet version is 1.0-beta23, Matlab version is R2015b. I'm not sure whether the version of MatConvNet matters, and what should I do to run the demo, any suggestion is appreciated!

'disableDropout' argument inside mdnet_features_convX.m

Hi, I have tried to run the demo on my computer. Below is my enviroment.

Ubuntu 14.04 LTS
Memory 15.6 GB
CPU: Intel® Core™ i7-6700K @ 4.00GHz × 8
Graphics: GeForce GTX TITAN X/PCIe/SSE2

After I compiled and installed the newest version of MatConvNet, I follow the instructions to run the demo. However, I come across the error below.

Error using vl_argparse (line 99)
Unknown parameter 'disableDropout'

Error in vl_simplenn (line 239)
opts = vl_argparse(opts, varargin);

Error in mdnet_features_convX (line 22)
res = vl_simplenn(net, batch, [], [], ...

Error in mdnet_run (line 37)
feat_conv = mdnet_features_convX(net_conv, img, pos_examples, opts);

Error in demo_tracking (line 22)
result = mdnet_run(conf.imgList, conf.gt(1,:), net);

Error in run (line 96)
evalin('caller', [script ';']);

I think this is due to the 'disableDropout' argument in line 23 of mdnet_features_convX.m where vl_simplenn of MatConvNet does not handle this argument ( both the newest version of MatConvnet and the one attached in this repository, below is line 227-236 of vl_simplenn where opts do not contain 'disableDropout' ).
I think this argument may be used to disable the default dropout setting of an older version of MatConvNet. But I am not quite sure. What is this argument for and what is the version of MatConvNet implementing MDNet? Is there something I didn't notice? Thank you!

opts.conserveMemory = false ;
opts.sync = false ;
opts.mode = 'normal' ;
opts.accumulate = false ;
opts.cudnn = true ;
opts.backPropDepth = +inf ;
opts.skipForward = false ;
opts.parameterServer = [] ;
opts.holdOn = false ;
opts = vl_argparse(opts, varargin);

the 'demo_tracking.m' aborted when runs hundreds of frames

Processing frame 137/315... 1.266084 seconds
Processing frame 138/315... 1.483625 seconds
Processing frame 139/315... Warning: A worker aborted during execution of the parfor loop. The parfor loop will
now run again on the remaining workers.

In parallel_function at 596
In mdnet_extract_regions at 17
In mdnet_features_convX at 9
In mdnet_run at 166
In demo_tracking at 22
In run at 63
Error using parallel_function (line 604)
All workers aborted during execution of the parfor loop.

Error in mdnet_extract_regions (line 17)
parfor i = 1:num_boxes

Error in mdnet_features_convX (line 9)
ims = mdnet_extract_regions(img, boxes, opts);

Error in mdnet_run (line 166)
feat_conv = mdnet_features_convX(net_conv, img, examples, opts);

Error in demo_tracking (line 22)
result = mdnet_run(conf.imgList, conf.gt(1,:), net);

Error in run (line 63)
evalin('caller', [script ';']);

The client lost connection to worker 1. This might be due to network problems, or
the interactive communicating job might have errored.

Using custom dataset with MDNet

Hi @HyeonseobNam,
I was trying to tweak the MDNet code with another custom dataset, which is VT77 dataset (http://cmp.felk.cvut.cz/~vojirtom/dataset/). I see in the training process, that MDNet use the sequence from multiple tracking sequences and needs to generate the sequence list first in pretraining/seqList. What is this list used for and how to generate this by using another dataset?

Thanks in advance

Is there a way around to compile matconvnet (1.0.beta10) on MATLAB 2017b on MacOS 14

I get the error below on compiling matconvnet on MacOS Mojave. I understand the error but have no clue on how to go around it.

warning: include path for stdlibc++ headers not found; pass '-std=libc++' on the command line to use the
libc++ standard library instead [-Wstdlibcxx-not-found]
In file included from
MDNet/matconvnet/matlab/src/bits/data.cpp:4:
In file included from
MDNet/matconvnet/matlab/src/bits/data.cu:13:
MDNet/matconvnet/matlab/src/bits/data.hpp:16:10:
fatal error: 'cstddef' file not found

Thank you!

Should the model be updated in the real world?

Thank you for your great work. I notice that the model would be updated if I want to predict the bounding box in a new video sequence. However, this seems to be impractical because it might be impossible to retrain the model in the client. Do you have any suggestion for this? Thank you again.

SRE and TRE results on OTB100

Hi @HyeonseobNam ,
Thank you for your brilliant work.
Recently when I tried to evaluate MDNet on OTB100 in detail, it seemed that only OPE results are available on your webpage. Therefore, I wonder if there is any chance to get SRE and TRE results of MDNet from your completed work?
Best Wishes!

error while running 'compile_mathconvnet'

Hi,

I'm getting this error while running compile_mathconvnet script in windows 7, using matlab r2016b with mingw as mex compiler:

>> compile_matconvnet
Starting parallel pool (parpool) using the 'local' profile ... connected to 2 workers.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
In file included from X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cpp:4:0:
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu: In member function 'vl::Error vl::MexTensor::init(vl::Device, const vl::TensorGeometry&)':
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu:177:43: warning: narrowing conversion of '(& geom)->vl::TensorGeometry::getHeight()' from 'vl::index_t {aka int}' to 'mwSize {aka long long unsigned int}' inside { } [-Wnarrowing]
   mwSize dimensions [4] = {geom.getHeight(),
                                           ^
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu:178:42: warning: narrowing conversion of '(& geom)->vl::TensorGeometry::getWidth()' from 'vl::index_t {aka int}' to 'mwSize {aka long long unsigned int}' inside { } [-Wnarrowing]
                            geom.getWidth(),
                                          ^
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu:179:42: warning: narrowing conversion of '(& geom)->vl::TensorGeometry::getDepth()' from 'vl::index_t {aka int}' to 'mwSize {aka long long unsigned int}' inside { } [-Wnarrowing]
                            geom.getDepth(),
                                          ^
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu:180:41: warning: narrowing conversion of '(& geom)->vl::TensorGeometry::getSize()' from 'vl::index_t {aka int}' to 'mwSize {aka long long unsigned int}' inside { } [-Wnarrowing]
                            geom.getSize()} ;
                                         ^
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu: In member function 'vl::Error vl::MexTensor::initWithZeros(vl::Device, const vl::TensorGeometry&)':
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu:233:43: warning: narrowing conversion of '(& geom)->vl::TensorGeometry::getHeight()' from 'vl::index_t {aka int}' to 'mwSize {aka long long unsigned int}' inside { } [-Wnarrowing]
   mwSize dimensions [4] = {geom.getHeight(),
                                           ^
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu:234:42: warning: narrowing conversion of '(& geom)->vl::TensorGeometry::getWidth()' from 'vl::index_t {aka int}' to 'mwSize {aka long long unsigned int}' inside { } [-Wnarrowing]
                            geom.getWidth(),
                                          ^
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu:235:42: warning: narrowing conversion of '(& geom)->vl::TensorGeometry::getDepth()' from 'vl::index_t {aka int}' to 'mwSize {aka long long unsigned int}' inside { } [-Wnarrowing]
                            geom.getDepth(),
                                          ^
X:\github\MDNet\matconvnet\matlab\src\bits\datamex.cu:236:41: warning: narrowing conversion of '(& geom)->vl::TensorGeometry::getSize()' from 'vl::index_t {aka int}' to 'mwSize {aka long long unsigned int}' inside { } [-Wnarrowing]
                            geom.getSize()} ;
                                         ^

MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
MEX completed successfully.
Building with 'MinGW64 Compiler (C++)'.
Error using vl_compilenn>mex_compile (line 411)
X:\github\MDNet\matconvnet\matlab\src\bits\impl\imread_gdiplus.cpp: In member function 'vl::Image
vl::ImageReader::Impl::read(const char*, float*)':
X:\github\MDNet\matconvnet\matlab\src\bits\impl\imread_gdiplus.cpp:91:83: error: 'mbstowcs_s' was not
declared in this scope
   mbstowcs_s(&n, filenamew, sizeof(filenamew)/sizeof(wchar_t), filename, _TRUNCATE);
                                                                                   ^
X:\github\MDNet\matconvnet\matlab\src\bits\impl\imread_gdiplus.cpp: In member function 'vl::Image
vl::ImageReader::Impl::readDimensions(const char*)':
X:\github\MDNet\matconvnet\matlab\src\bits\impl\imread_gdiplus.cpp:149:83: error: 'mbstowcs_s' was
not declared in this scope
   mbstowcs_s(&n, filenamew, sizeof(filenamew)/sizeof(wchar_t), filename, _TRUNCATE);
                                                                                   ^


Error in vl_compilenn (line 371)
parfor i = 1:numel(horzcat(lib_src, mex_src))

Error in compile_matconvnet (line 13)
vl_compilenn;

Error using vl_nnconv

Kindly help me, to solve the following error,

Error using vl_nnconv
An input is not a numeric array (or GPU support not compiled).

Error in vl_simplenn (line 153)
res(i+1).x = vl_nnconv(res(i).x, l.filters, l.biases, 'pad', l.pad, 'stride', l.stride) ;

Error in mdnet_features_convX (line 20)
res = vl_simplenn(net, batch, [], [], ...

Error in mdnet_run (line 37)
feat_conv = mdnet_features_convX(net_conv, img, pos_examples, opts);

Error in demo_tracking (line 22)
result = mdnet_run(conf.imgList, conf.gt(1,:), net);

I want to use the new model ,but it got something wrong.

@HyeonseobNam
Thanks for your open source work.
Recently ,I only used the two layer's CNN from VGG, then i got the 5*5 feature map,but when I ran the demo_tracking.m with using my pretrained mode(the training phase was OK)l, it get something wrong like issue 5.but the error log says something wrong with the mdnet_finetune_hnm.m file. So I want to modidified the mdnet_finetune_hnm.m to adjust to my model. Thanks.

error compiling

I am attempting to compile MDNet, and I am getting the following error:

Error using vl_compilenn>mex_compile (line
411)
In file included from
C:\Users\MDNet-master\matconvnet\matlab\src\bits\impl\imread_helpers.hpp:17:0,
                 from
C:\Users\MDNet-master\matconvnet\matlab\src\bits\impl\imread_gdiplus.cpp:14:
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_hadd_epi16(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:42:70:
error: '__builtin_ia32_phaddw128' was not
declared in this scope
   return (__m128i) __builtin_ia32_phaddw128
   ((__v8hi)__X, (__v8hi)__Y);
                                                                      ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_hadd_epi32(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:48:70:
error: '__builtin_ia32_phaddd128' was not
declared in this scope
   return (__m128i) __builtin_ia32_phaddd128
   ((__v4si)__X, (__v4si)__Y);
                                                                      ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_hadds_epi16(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:54:71:
error: '__builtin_ia32_phaddsw128' was not
declared in this scope
   return (__m128i) __builtin_ia32_phaddsw128
   ((__v8hi)__X, (__v8hi)__Y);
                                                                       ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_hadd_pi16(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:60:65:
error: '__builtin_ia32_phaddw' was not
declared in this scope
   return (__m64) __builtin_ia32_phaddw
   ((__v4hi)__X, (__v4hi)__Y);
                                                                 ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_hadd_pi32(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:66:65:
error: '__builtin_ia32_phaddd' was not
declared in this scope
   return (__m64) __builtin_ia32_phaddd
   ((__v2si)__X, (__v2si)__Y);
                                                                 ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_hadds_pi16(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:72:66:
error: '__builtin_ia32_phaddsw' was not
declared in this scope
   return (__m64) __builtin_ia32_phaddsw
   ((__v4hi)__X, (__v4hi)__Y);
                                                                  ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_hsub_epi16(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:78:70:
error: '__builtin_ia32_phsubw128' was not
declared in this scope
   return (__m128i) __builtin_ia32_phsubw128
   ((__v8hi)__X, (__v8hi)__Y);
                                                                      ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_hsub_epi32(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:84:70:
error: '__builtin_ia32_phsubd128' was not
declared in this scope
   return (__m128i) __builtin_ia32_phsubd128
   ((__v4si)__X, (__v4si)__Y);
                                                                      ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_hsubs_epi16(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:90:71:
error: '__builtin_ia32_phsubsw128' was not
declared in this scope
   return (__m128i) __builtin_ia32_phsubsw128
   ((__v8hi)__X, (__v8hi)__Y);
                                                                       ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_hsub_pi16(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:96:65:
error: '__builtin_ia32_phsubw' was not
declared in this scope
   return (__m64) __builtin_ia32_phsubw
   ((__v4hi)__X, (__v4hi)__Y);
                                                                 ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_hsub_pi32(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:102:65:
error: '__builtin_ia32_phsubd' was not
declared in this scope
   return (__m64) __builtin_ia32_phsubd
   ((__v2si)__X, (__v2si)__Y);
                                                                 ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_hsubs_pi16(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:108:66:
error: '__builtin_ia32_phsubsw' was not
declared in this scope
   return (__m64) __builtin_ia32_phsubsw
   ((__v4hi)__X, (__v4hi)__Y);
                                                                  ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i
_mm_maddubs_epi16(__m128i, __m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:114:75:
error: '__builtin_ia32_pmaddubsw128' was not
declared in this scope
   return (__m128i)
   __builtin_ia32_pmaddubsw128 ((__v16qi)__X,
   (__v16qi)__Y);
                                                                           ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_maddubs_pi16(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:120:68:
error: '__builtin_ia32_pmaddubsw' was not
declared in this scope
   return (__m64) __builtin_ia32_pmaddubsw
   ((__v8qi)__X, (__v8qi)__Y);
                                                                    ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i
_mm_mulhrs_epi16(__m128i, __m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:126:72:
error: '__builtin_ia32_pmulhrsw128' was not
declared in this scope
   return (__m128i)
   __builtin_ia32_pmulhrsw128 ((__v8hi)__X,
   (__v8hi)__Y);
                                                                        ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_mulhrs_pi16(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:132:67:
error: '__builtin_ia32_pmulhrsw' was not
declared in this scope
   return (__m64) __builtin_ia32_pmulhrsw
   ((__v4hi)__X, (__v4hi)__Y);
                                                                   ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i
_mm_shuffle_epi8(__m128i, __m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:138:72:
error: '__builtin_ia32_pshufb128' was not
declared in this scope
   return (__m128i) __builtin_ia32_pshufb128
   ((__v16qi)__X, (__v16qi)__Y);
                                                                        ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_shuffle_pi8(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:144:65:
error: '__builtin_ia32_pshufb' was not
declared in this scope
   return (__m64) __builtin_ia32_pshufb
   ((__v8qi)__X, (__v8qi)__Y);
                                                                 ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_sign_epi8(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:150:72:
error: '__builtin_ia32_psignb128' was not
declared in this scope
   return (__m128i) __builtin_ia32_psignb128
   ((__v16qi)__X, (__v16qi)__Y);
                                                                        ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_sign_epi16(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:156:70:
error: '__builtin_ia32_psignw128' was not
declared in this scope
   return (__m128i) __builtin_ia32_psignw128
   ((__v8hi)__X, (__v8hi)__Y);
                                                                      ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_sign_epi32(__m128i,
__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:162:70:
error: '__builtin_ia32_psignd128' was not
declared in this scope
   return (__m128i) __builtin_ia32_psignd128
   ((__v4si)__X, (__v4si)__Y);
                                                                      ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_sign_pi8(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:168:65:
error: '__builtin_ia32_psignb' was not
declared in this scope
   return (__m64) __builtin_ia32_psignb
   ((__v8qi)__X, (__v8qi)__Y);
                                                                 ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_sign_pi16(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:174:65:
error: '__builtin_ia32_psignw' was not
declared in this scope
   return (__m64) __builtin_ia32_psignw
   ((__v4hi)__X, (__v4hi)__Y);
                                                                 ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_sign_pi32(__m64,
__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:180:65:
error: '__builtin_ia32_psignd' was not
declared in this scope
   return (__m64) __builtin_ia32_psignd
   ((__v2si)__X, (__v2si)__Y);
                                                                 ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_alignr_epi8(__m128i,
__m128i, int)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:188:32:
error: '__builtin_ia32_palignr128' was not
declared in this scope
            (__v2di)__Y, __N * 8);
                                ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_alignr_pi8(__m64,
__m64, int)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:195:27:
error: '__builtin_ia32_palignr' was not
declared in this scope
       (__v1di)__Y, __N * 8);
                           ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_abs_epi8(__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:211:57:
error: '__builtin_ia32_pabsb128' was not
declared in this scope
   return (__m128i) __builtin_ia32_pabsb128
   ((__v16qi)__X);
                                                         ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_abs_epi16(__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:217:56:
error: '__builtin_ia32_pabsw128' was not
declared in this scope
   return (__m128i) __builtin_ia32_pabsw128
   ((__v8hi)__X);
                                                        ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m128i _mm_abs_epi32(__m128i)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:223:56:
error: '__builtin_ia32_pabsd128' was not
declared in this scope
   return (__m128i) __builtin_ia32_pabsd128
   ((__v4si)__X);
                                                        ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_abs_pi8(__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:229:51:
error: '__builtin_ia32_pabsb' was not
declared in this scope
   return (__m64) __builtin_ia32_pabsb
   ((__v8qi)__X);
                                                   ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_abs_pi16(__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:235:51:
error: '__builtin_ia32_pabsw' was not
declared in this scope
   return (__m64) __builtin_ia32_pabsw
   ((__v4hi)__X);
                                                   ^
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:
In function '__m64 _mm_abs_pi32(__m64)':
C:/ProgramData/MATLAB/SupportPackages/R2019a/3P.instrset/mingw_w64.instrset/lib/gcc/x86_64-w64-mingw32/6.3.0/include/tmmintrin.h:241:51:
error: '__builtin_ia32_pabsd' was not
declared in this scope
   return (__m64) __builtin_ia32_pabsd
   ((__v2si)__X);
                                                   ^


Error in vl_compilenn (line 371)
parfor i = 1:numel(horzcat(lib_src, mex_src))

Error in compile_matconvnet (line 10)
vl_compilenn;

I am using Windows 10, and I installed mingw via Matlab's add-ons.

what shoud I do if I use CUDA8.0?

Thanks for ur excellent works!
Recently I run the demo_tracking.m, but when I use compile vl_compilenn('enableGpu', true) based on origin matconvnet(your matonvnet), it occurs that

, in other words, old version matconvnet can not match new CUDA version, so what should I do if I want to use GPU?

Running the MDNet demo on CPU

Hi,

How to run the model on CPU?

I tried your instruction
You can also disable the GPU support by setting opts.useGpu in 'tracking/mdnet_init.m' to false (not recommended).

But, still I am getting error in running demo_tracking

Trial>> demo_tracking
Initialization...
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Warning: While loading an object of class 'gpuArray':
There is a problem with the CUDA driver or with this GPU device. Be sure that you have a supported GPU and that the latest driver
is installed. 
> In mdnet_init (line 68)
  In mdnet_run (line 29)
  In demo_tracking (line 22) 
Error using gpuArray/gather
Attempting to access the property or method of an invalid object.

Error in vl_simplenn_move>@(x)gather(x) (line 17)
  case 'cpu', moveop = @(x) gather(x) ;

Error in vl_simplenn_move (line 26)
          net.layers{l}.(f) = moveop(net.layers{l}.(f)) ;

Error in mdnet_init (line 89)
    net_conv = vl_simplenn_move(net_conv, 'cpu') ;

Error in mdnet_run (line 29)
[net_conv, net_fc, opts] = mdnet_init(img, net);

Error in demo_tracking (line 22)
result = mdnet_run(conf.imgList, conf.gt(1,:), net);

Is there C++ or Caffe version MDNet ?

Asking for help about evaluating mdnet on VOT dataset

What's the fps of MDNet using GPU?

hi @HyeonseobNam ,
Would you please to tell the fps you achieved in your test environment?
The following is my testcase. Is that closed to yours?

Initialization...
extract features...
training cnn...
im2col: CUDA kernel error (invalid texture reference)
Processing frame 2/215... 2.725772 seconds
Processing frame 3/215... 2.145009 seconds
Processing frame 4/215... 1.917557 seconds
Processing frame 5/215... 2.628119 seconds
Processing frame 6/215... 2.196728 seconds
Processing frame 7/215... 1.892060 seconds
Processing frame 8/215... 2.746299 seconds
Processing frame 9/215... 2.051056 seconds
Processing frame 10/215...

A issue in operate the code to start the demo of MD Net

First,I run the file named setup_mdnet.m in matlab ,then run the compile_matconvnet.m similarly in the matlab and at last when I run the file named demo_tracking.m,the issue is occuring.

Environment:

system:Ubuntu16.04,matlab:R2016b

Issues as fellows:

>> demo_tracking
Initialization...
Invalid MEX-file '/home/linhan/Desktop/MD/MDNet-master/matconvnet/matlab/mex/vl_nnconv.mexa64':
/usr/local/MATLAB/R2016b/bin/glnxa64/../../sys/os/glnxa64/libstdc++.so.6: version `GLIBCXX_3.4.21' not found (required by
/home/linhan/Desktop/MD/MDNet-master/matconvnet/matlab/mex/vl_nnconv.mexa64).

Error in vl_simplenn (line 153)
      res(i+1).x = vl_nnconv(res(i).x, l.filters, l.biases, 'pad', l.pad, 'stride', l.stride) ;

Error in mdnet_features_convX (line 20)
    res = vl_simplenn(net, batch, [], [], ...

Error in mdnet_run (line 37)
    feat_conv = mdnet_features_convX(net_conv, img, pos_examples, opts);

Error in demo_tracking (line 22)
result = mdnet_run(conf.imgList, conf.gt(1,:), net);

What the issues is?This code I run in this environment is OK long time ago,but I operate it today ,the issues is occuring.

evaluation code

Hi @HyeonseobNam

is there a code for measuring the accuracy and robutsness of the tracker ?. Thanks.