eldar/deepcut

Multi Person Pose Estimation

deepcut's Issues

GPU out of memory error

Hello everyone,
I'm testing your code for a project and I'm getting some of the following errors when I run demo_multiperson.m:

(in the console)
F0923 15:47:45.735599 4029 syncedmem.cpp:56] Check failed: error == cudaSuccess (2 vs. 0) out of memory *** Check failure stack trace: *** Killed

(in matlab)
Cleared 0 solvers and 0 stand-alone nets save dir .../git/deepcut/data/mpii-multiperson/scoremaps/test testing from net file /home/marco/Desktop/mauro-skeletal-tracker/git/deepcut/data/caffe-models/ResNet-101-mpii-multiperson.caffemodel deepcut: test (MPII multiperson test) 2/1758

with a Matlab crash.

My video card is an NVidia GeForce GTX 760 Ti 2GB.

I'm new to deep learning and Caffe, but I've read on the web that sometimes is possible to run tests on less memory capable graphics cards, as in my case, by changing some parameters (like the batch-size). Is that possible in deepcut? Where could I change those parameters in your code?
Thank you in advance!

Unable to compile external/solver

After successfully installing Gurobi 6.0.5 I try to make the solver. During linking I get the following:

CMakeFiles/solver-callback.dir/src/pose/research/solver-callback.cxx.o: In function void andres::ilp::Gurobi<double>::addConstraint<unsigned long*, double const*>(unsigned long*, unsigned long*, double const*, double, double) [clone .constprop.252]': solver-callback.cxx:(.text+0x2fc): undefined reference toGRBModel::addConstr(GRBLinExpr const&, char, GRBLinExpr const&, std::__cxx11::basic_string<char, std::char_traits, std::allocator >)'
solver-callback.cxx:(.text+0x363): undefined reference to GRBModel::addConstr(GRBLinExpr const&, char, GRBLinExpr const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)' CMakeFiles/solver-callback.dir/src/pose/research/solver-callback.cxx.o: In functionandres::ilp::Gurobi::Callback::callback()':
solver-callback.cxx:(.text._ZN6andres3ilp6GurobiIdE8Callback8callbackEv[_ZN6andres3ilp6GurobiIdE8Callback8callbackEv]+0xbf): undefined reference to GRBException::getMessage[abi:cxx11]() const' CMakeFiles/solver-callback.dir/src/pose/research/solver-callback.cxx.o: In functionvoid andres::ilp::Gurobi::addConstraint<__gnu_cxx::__normal_iterator<unsigned long*, std::vector<unsigned long, std::allocator > >, __gnu_cxx::__normal_iterator<double*, std::vector<double, std::allocator > > >(__gnu_cxx::__normal_iterator<unsigned long*, std::vector<unsigned long, std::allocator > >, __gnu_cxx::__normal_iterator<unsigned long*, std::vector<unsigned long, std::allocator > >, gnu_cxx::normal_iterator<double*, std::vector<double, std::allocator > >, double, double) [clone .constprop.251]':
solver-callback.cxx:(.text.ZN6andres3ilp6GurobiIdE13addConstraintIN9__gnu_cxx17__normal_iteratorIPmSt6vectorImSaImEEEENS5_IPdS7_IdSaIdEEEEEEvT_SF_T0_dd.constprop.251[ZN4pose13PoseEstimatorIN6andres3ilp6GurobiIdEE24FeasibleSolutionCallbackE25addAllCouplingConstraintsEv]+0xb2): undefined reference to GRBModel::addConstr(GRBLinExpr const&, char, GRBLinExpr const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)' CMakeFiles/solver-callback.dir/src/pose/research/solver-callback.cxx.o: In functionvoid andres::ilp::Gurobi::addConstraint<unsigned long*, double const*>(unsigned long, unsigned long, double const, double, double) [clone .constprop.253]':
solver-callback.cxx:(.text.ZN6andres3ilp6GurobiIdE13addConstraintIPmPKdEEvT_S7_T0_dd.constprop.253[ZN4pose13PoseEstimatorIN6andres3ilp6GurobiIdEE24FeasibleSolutionCallbackE36addAllImpossiblePartClassConstraintsEv]+0xb3): undefined reference to GRBModel::addConstr(GRBLinExpr const&, char, GRBLinExpr const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)' CMakeFiles/solver-callback.dir/src/pose/research/solver-callback.cxx.o: In functionvoid andres::ilp::Gurobi::addConstraint<unsigned long, double const>(unsigned long, unsigned long, double const, double, double)':
solver-callback.cxx:(.text._ZN6andres3ilp6GurobiIdE13addConstraintIPmPKdEEvT_S7_T0_dd[_ZN6andres3ilp6GurobiIdE13addConstraintIPmPKdEEvT_S7_T0_dd]+0x13c): undefined reference to GRBModel::addConstr(GRBLinExpr const&, char, GRBLinExpr const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)' solver-callback.cxx:(.text._ZN6andres3ilp6GurobiIdE13addConstraintIPmPKdEEvT_S7_T0_dd[_ZN6andres3ilp6GurobiIdE13addConstraintIPmPKdEEvT_S7_T0_dd]+0x19f): undefined reference toGRBModel::addConstr(GRBLinExpr const&, char, GRBLinExpr const&, std::__cxx11::basic_string<char, std::char_traits, std::allocator >)'
solver-callback.cxx:(.text._ZN6andres3ilp6GurobiIdE13addConstraintIPmPKdEEvT_S7_T0_dd[_ZN6andres3ilp6GurobiIdE13addConstraintIPmPKdEEvT_S7_T0_dd]+0x1f3): undefined reference to GRBModel::addConstr(GRBLinExpr const&, char, GRBLinExpr const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)' CMakeFiles/solver-callback.dir/src/pose/research/solver-callback.cxx.o:solver-callback.cxx:(.text._ZN6andres3ilp6GurobiIdE13addConstraintIN9__gnu_cxx17__normal_iteratorIPmSt6vectorImSaImEEEENS5_IPKmS9_EEEEvT_SE_T0_dd[_ZN6andres3ilp6GurobiIdE13addConstraintIN9__gnu_cxx17__normal_iteratorIPmSt6vectorImSaImEEEENS5_IPKmS9_EEEEvT_SE_T0_dd]+0x169): more undefined references toGRBModel::addConstr(GRBLinExpr const&, char, GRBLinExpr const&, std::__cxx11::basic_string<char, std::char_traits, std::allocator >)' follow
CMakeFiles/solver-callback.dir/src/pose/research/solver-callback.cxx.o: In function main': solver-callback.cxx:(.text.startup+0x5f9): undefined reference toGRBException::getMessageabi:cxx11 const'
collect2: error: ld returned 1 exit status
CMakeFiles/solver-callback.dir/build.make:102: recipe for target 'solver-callback' failed
make[3]: *** [solver-callback] Error 1
CMakeFiles/Makefile2:141: recipe for target 'CMakeFiles/solver-callback.dir/all' failed
make[2]: *** [CMakeFiles/solver-callback.dir/all] Error 2
CMakeFiles/Makefile2:153: recipe for target 'CMakeFiles/solver-callback.dir/rule' failed
make[1]: *** [CMakeFiles/solver-callback.dir/rule] Error 2
Makefile:155: recipe for target 'solver-callback' failed
make: *** [solver-callback] Error 2

Gurobi environment set to:
GUROBI_HOME=/opt/gurobi605/linux64
LD_LIBRARY_PATH=/opt/gurobi605/linux64/lib

Output from CMake:
coert@beta039608 /mnt/33EE47FC4F4F9C75/Install/deepcut_multiperson/external/solver $ cmake . -DGUROBI_ROOT_DIR=/opt/gurobi605/linux64 -DGUROBI_VERSION=60
-- Try OpenMP C flag = [-fopenmp]
-- Performing Test OpenMP_FLAG_DETECTED
-- Performing Test OpenMP_FLAG_DETECTED - Success
-- Try OpenMP CXX flag = [-fopenmp]
-- Performing Test OpenMP_FLAG_DETECTED
-- Performing Test OpenMP_FLAG_DETECTED - Success
-- Found OpenMP: -fopenmp
-- OpenMP found
-- Found HDF5: /usr/lib/x86_64-linux-gnu/hdf5/serial/lib/libhdf5.so;/usr/lib/x86_64-linux-gnu/libpthread.so;/usr/lib/x86_64-linux-gnu/libsz.so;/usr/lib/x86_64-linux-gnu/libz.so;/usr/lib/x86_64-linux-gnu/libdl.so;/usr/lib/x86_64-linux-gnu/libm.so (found version "1.8.16")
-- Found GUROBI: /opt/gurobi605/linux64/lib/libgurobi60.so
-- Found GLUT: /usr/lib/x86_64-linux-gnu/libglut.so
-- Configuring done
-- Generating done
-- Build files have been written to: /mnt/33EE47FC4F4F9C75/Install/deepcut_multiperson/external/solver

What am I missing here?

Error in _cutoff_tile function

Hello!

I get the following error in the _cutoff_tile function:
deepcut-cnn/python/pose/estimate_pose.py in _cutoff_tile(sm, num_tiles, idx, cut_off, is_x)
259 sm = sm[cut_off:, ...]
260 else:
--> 261 sm = sm[cut_off:-cut_off, ...]
262 if is_x:
263 sm = sm.transpose((1, 0, 2, 3))

TypeError: slice indices must be integers or None or have an index method

From what I have seen in the estimate_pose.py file,
cut_off = rf / stride
and rf = 224 and _STRIDE = 8. (with the . (dot)), so cut_off is indeed a float number. I tried to change this, but then it does not work correctly.

And this does not happen with all the pictures, and I couldn't find a pattern in the picture where it does.
Any help?

Thank you!

Failed to build Cafee

I ran make -j8 all matcafee and received error src/caffe/layers/softmax_loss_vec_layer.cpp:254:10: error: redefinition of ‘void caffe::SoftmaxWithLossVecLayer<Dtype>::Forward_gpu(const std::vector<caffe::Blob<Dtype>*>&, const std::vector<caffe::Blob<Dtype>*>&)’. Googling does not give any result. Please note that due to my limited GPU, I had to build using only CPU.

Any help is much appreciated.

Runtime error

Could you please let me know the issue with my demo?

error.txt
...
I1016 22:46:16.365223 24943 net.cpp:816] Ignoring source layer loss_loc
I1016 22:46:16.374922 24943 net.cpp:816] Ignoring source layer loss_next
save dir /gpfs/work/f/fuf111/deepcut/data/mpii-multiperson/scoremaps/test
testing from net file /gpfs/work/f/fuf111/deepcut/data/caffe-models/ResNet-101-mpii-multiperson.caffemodel
deepcut: test (MPII multiperson test) 2/1758
F1016 22:46:17.488354 24943 syncedmem.cpp:136] Cannot use GPU in CPU-only Caffe: check mode.
*** Check failure stack trace: ***

Evaluation on MPII Human Pose Dataset

Hi，I want to use the evaluation script evaluateAP.m downloaded on MPII Human Pose Dataset to measure my model's performance, but I could not find the annolist_dataset_v12.mat file loaded in this line of the script.

load([p.gtDir '/annolist_dataset_v12'],'annolist');

Do you know how to get this file, thanks.

Requirements

Hello!

I would like to know what the minimum requirements on the os are, such as cpu, ram, graphics card...

Thanks!

Error in demo_multiperson (line 12)

ERROR:

calling solver: /home/yulijun/deepcut/external/solver/solver-callback /home/yulijun/deepcut/data//mpii-multiperson/multicut//problem-0002.h5 /home/yulijun/deepcut/data//mpii-multiperson/multicut//solution-0002.h5 m 86400
echo $HOSTNAME: Aborted
hostname:

malloc: unknown:0: assertion botched

free: called with unallocated block argument
last command: (null)
Aborting...

malloc: unknown:0: assertion botched

free: called with unallocated block argument
last command: (null)
Aborting...export GRB_LICENSE_FILE=/home/yulijun/deepcut/data/gurobi.lic; /home/yulijun/deepcut/external/solver/solver-callback /home/yulijun/deepcut/data//mpii-multiperson/multicut//problem-0002.h5 /home/yulijun/deepcut/data//mpii-multiperson/multicut//solution-0002.h5 m 86400: Aborted
Elapsed time is 0.099782 seconds.
Error using test_spatial_app_neighbour (line 444)
solver error

Error in demo_multiperson (line 12)
test_spatial_app_neighbour(experiment_index, image_index, 1, true, true);

hope you give me some advice!

solutionFname

@eldar
hello, i am interested in your work of multiperson. here i met one problem in test_apatial_app_neighhbour.m: in line 453
unLab = marray_load(solutionFname, dataName);
and i find that the solutionFname desn't exit. i see your code and find that you really don't save solutionFname. i think something is missing or the code has some bug.Do you think so?

Could NOT find GLUT?

When i was building solver,i got a question that could not find glut.Do you know how to solve the problem?Thanks.

how to computer the mean average precision ?

Hi,I try to drop the net resolution to raise the fps. and then I use the MPII Human Pose Dataset mentioned in your article to check out how much the map drop.On the Website:
http://human-pose.mpi-inf.mpg.de/#evaluation I mess around by the README.

pos = zeros(length(rect),2);
for ridx = 1:length(rect)
    pos(ridx,:) = [rect(ridx).objpos.x rect(ridx).objpos.y];
end
x1 = min(pos(:,1)); y1 = min(pos(:,2)); x2 = max(pos(:,1)); y2 = max(pos(:,2));

where can I get the rect ?

load('groups_v12.mat','groups');
[imgidxs_multi_test,rectidxs_multi_test] = getMultiPersonGroups(groups,RELEASE,false);

what is the groups_v12.mat ?
I search on the web for a day, and get little effective information. can you give me some suggestion ? thx

Error using hdf5writec

Hi. i succesfully installed HDF5 1.8 CMake C++ 11 CUDA 8.0 matlab2014a and made the project successfully, but there are some problems. The first is i can't open matlab using ./start_matlab.sh, instead i open matlab in the path of my matlab installation , and run sudo ./matlab.sh under bin. i don't know whether it matters??? when i tried the models: startup and demo_multiperson, it occured some problems......
Cleared 0 solvers and 0 stand-alone nets
save dir /home/yulijun/PoseTrack-CVPR2017/deepcut/data/mpii-multiperson/scoremaps/test
testing from net file /home/yulijun/PoseTrack-CVPR2017/deepcut/data/caffe-models/ResNet-101-mpii-multiperson.caffemodel
deepcut: test (MPII multiperson test) 2/1758
Cleared 0 solvers and 1 stand-alone nets
test_spatial_hist()
expidx: 1
firstidx: 2
nImgs: 1
/bin/bash: /home/yulijun/Downloads/MATLAB/bin/glnxa64/libtiff.so.5: no version information available (required by /usr/lib/x86_64-linux-gnu/libopencv_highgui.so.2.4)

malloc: unknown:0: assertion botched

free: called with unallocated block argument
last command: (null)
Aborting...mkdir -p /home/yulijun/PoseTrack-CVPR2017/deepcut/data//mpii-multiperson/multicut/: Aborted
multicutDir: /home/yulijun/PoseTrack-CVPR2017/deepcut/data//mpii-multiperson/multicut/
/bin/bash: /home/yulijun/Downloads/MATLAB/bin/glnxa64/libtiff.so.5: no version information available (required by /usr/lib/x86_64-linux-gnu/libopencv_highgui.so.2.4)

malloc: unknown:0: assertion botched

free: called with unallocated block argument
last command: (null)
Aborting...mkdir -p /home/yulijun/PoseTrack-CVPR2017/deepcut/data/mpii-multiperson/multicut/vis: Aborted
Loading spatial model from /home/yulijun/PoseTrack-CVPR2017/deepcut/data/pairwise
recompute 1
imgidx: 2
loading features... done!
im_name im00044_1
Warning: Image is too big to fit on screen; displaying at 67%

In imuitools/private/initSize at 71
In imshow at 282
In test_spatial_app_neighbour at 210
In demo_multiperson at 12
number of detections before 1018 after 1018
crop rectangle (x, y, width, height): (1 1 1519 681)
preserve 80/3960 detections
compute pairwise across parts prob
Accuracy = 94.5% (6048/6400)
Accuracy = 93.75% (6000/6400)
Accuracy = 75% (4800/6400)
Accuracy = 99.5% (6368/6400)
Accuracy = 79.75% (5104/6400)
Accuracy = 68% (4352/6400)
save problem
Error using hdf5writec
Call to HDF5 library failed (fileCreation): "createHDF5File: Couldn't create file.".

Error in hdf5write (line 138)
hdf5writec(file, details, varargin{dset_idx});

Error in marray_save (line 15)
hdf5write(file_name, dataset_name, M, 'WriteMode', write_mode);

Error in test_spatial_app_neighbour (line 395)
marray_save(problemFname, dataName, dets.unProb, write_mode);

Error in demo_multiperson (line 12)
test_spatial_app_neighbour(experiment_index, image_index, 1, true, true);

PS: i installed the hdf5 1.8.18 after building caffe and matlab, and solver.....

hope you give me your advice ,Thank you very much!

How to train?

Is there any possible way to train on my own dataset? Thank you!

solver-callback

Hi,
I want to test the matlab demo of the deep cut code ,
I have compiled caffe and deep cut code successfully.

but in execution of the matlab demo it ends up with error ,
since there is not any folder as solver-callback in the original code in external/solver/solver-callback.

What is the source of this error ?

Thanks.

Errors while compiling CPU-only version

I am facing below errors while trying to compile caffe with "CPU_ONLY := 1" flag. There is some error related to STUB_GPU macro. It leads to redefinition of all the functions which gives compiler error. Below is the screenshot attached:

ResNet-50.caffemodel

Due to some reasons, I need your ResNet-50.caffemodel, Can you supply it??