Aleksandr Rybnikov
09b73b2dc7
Blobs reuse improvement ( #1205 )
...
* Reuse deep learning output blobs
* Changed order for iterating through blobs while seeking memory. Refactored a little.
8 years ago
Aleksandr Rybnikov
692ba7bafd
Fixed SSD example, added test
8 years ago
dkurt
78ff9d931a
Import SoftMax, LogSoftMax layers from Torch
8 years ago
Dmitry Kurtaev
2abf45d1ee
Made some deep learning layers params are public ( #1134 )
8 years ago
Aleksandr Rybnikov
78fabfe606
Changed blobFromImages() function, so it can handle 1-channel images ( #1185 )
8 years ago
dkurt
d5ac902f18
Fix changes with allocating memory outside layers.
8 years ago
Vladislav Sovrasov
50a52424fa
dnn: fix possible build problem with tensor flow, get rid of catch(...)
8 years ago
Vladislav Sovrasov
10aee918c0
Fix DNN module build without torch
8 years ago
Aleksandr Rybnikov
a5d0ef520c
Added statistics functions
8 years ago
Vadim Pisarevsky
f78bb2c4e7
fixed shape() function ( #1177 )
8 years ago
Aleksandr Rybnikov
9b73fee29d
Made separate functions for computing output shapes for all layers. Removed output blobs allocation from layers
8 years ago
dkurt
eca606277a
Enabled parallel computation of ReLU activation layer
8 years ago
Aleksandr Rybnikov
08ade5e10e
Disabled FCN test on win32, changed googlenet test
8 years ago
dkurt
743f82c002
Removed an extra reallocation of networks every setBlob call.
8 years ago
Alexander Alekhin
ddc58e76ad
dnn: extra testdata files from OPENCV_DNN_TEST_DATA_PATH env
8 years ago
Aleksandr Rybnikov
3b01eaa3d8
Fixed bugs after dnn refactoring. Added new tests for layers ( #1142 )
...
* Fixed bugs after dnn refactoring. Added new tests for layers
* Fixed torch tests
8 years ago
Vadim Pisarevsky
7578908901
removed trailing whitespaces
8 years ago
Vadim Pisarevsky
9e26b24d04
improved speed of ENet processing.
8 years ago
Aleksandr Rybnikov
053303ab44
Added python scripts to estimating accuracy
8 years ago
Vladislav Sovrasov
96222f36c3
dnn: fix build without Torch
8 years ago
Vadim Pisarevsky
3f5b465501
refactored DNN ( #1102 )
...
* the first commit in the merged dnn: convert some public API from Blob's to Mat's
* temporarily or permantently removed OpenCL optimizations, which are not always stable nor usually very efficient; we'll likely use Halide instead
* got rid of Blob and BlobShape completely; use cv::Mat and std::vector<int> instead
* fixed a few compile errors
* got rid of separate .hpp files with layer declarations; instead, put everything into the respective .cpp files
* normalized all the layers' constructors; we concentrate on loading deep networks layers from files instead of constructing them from scratch, so we retained only SomeLayer::SomeLayer(const LayerParams& params); constructors
* fixed sample compilation
* suppress doxygen warnings
* trying to fix python bindings generation for DNN module
* temporarily disable python bindings while we refactor the module
* fix win32/win64 compile errors; remove trailing whitespaces
* fix win32/win64 compile errors; remove trailing whitespaces
8 years ago
Aleksandr Rybnikov
f3d751f44e
Fixed bug
8 years ago
Alexander Alekhin
cde15a7726
cmake: update ocv_download
8 years ago
Maksim Shabunin
95d432db5d
Download cache
8 years ago
Aleksandr Rybnikov
fd89b57453
Changed parameters of unpooling layer
8 years ago
Dmitry Kurtaev
8d74aaad3a
Merge pull request #1006 from dkurt:get_layer_inputs
...
Getting layer inputs (#1006 )
8 years ago
dkurt
d95fdc7df6
Elementwise layers parameters initialization.
...
They are initialized only in functors, not in inherited fields.
8 years ago
dkurt
a28b3f3086
Added missed layers names and types initializations
8 years ago
Alexander Alekhin
6b980ac59c
cmake: don't pass additional libraries as modules
8 years ago
arrybn
b964d3a18b
Added few tests for torch
8 years ago
Alexander Alekhin
e6550fcad5
dnn: fix OpenCL code in pooling_ocl
8 years ago
arrybn
0cef050337
Updated proto files, added layers to caffe
8 years ago
Aleksandr Rybnikov
9c6c69a4e4
Fixed and added tests ( #987 )
8 years ago
Vladislav Sovrasov
7c7c2848cf
Fix python bindings for dnn when Torch is disabled
8 years ago
Alexander Alekhin
4bf9cb5520
dnn: fix 'log1p' re-definition in THGeneral
8 years ago
Alexander Alekhin
62cc8d86e9
dnn: build fixes
8 years ago
Alexander Alekhin
2feb23cf36
cmake: include contrib modules into "opencv_world"
8 years ago
Alexander Alekhin
695996e826
dnn: fix iOS build warnings
8 years ago
Alexander Alekhin
3210f7848f
dnn: don't disable module on aarch64
8 years ago
Vladislav Sovrasov
620d74ab1f
Put Torch importer implementation under ifdef
8 years ago
arrybn
310a096b46
Added im2row, tiny optimiziations
8 years ago
Alexander Alekhin
1bd279e49d
fix lapack support
8 years ago
arrybn
e784f137f6
Added Torch ENet support
8 years ago
Vadim Pisarevsky
c84a6604c1
moved BLAS/LAPACK detection scripts to the main repository. Give preference to the external BLAS over OpenCV's gemm, which improves performance of dnn module by a lot
8 years ago
Alexander Alekhin
c82c887d10
cmake: unpack protobuf into binary directory
8 years ago
Alexander Alekhin
1da7edcd19
cmake: disable ICC warnings from protobuf code
8 years ago
Aleksandr Rybnikov
327aa89c37
Added TensorFlow models loading
8 years ago
Alexander Alekhin
b7ce1ec369
dnn: rebuild caffe.proto with protobuf 3.1.0
8 years ago
Alexander Alekhin
57d63db9c3
3rdparty: libprotobuf v3.1.0
8 years ago
Alexander Alekhin
f2af57c6d7
3rdparty: remove protobuf sources
8 years ago