Avi Haiat
23785d5342
Expose python binding for running multiple detectors at once ( #328 )
...
* Expose python binding for running multiple detectors at once
* Remove unwanted typedef
2016-11-12 07:54:48 -05:00
MambaWong
764e9185e7
Modify to support both train-images-idx3-ubyte and train-images.idx3-ubyte styles of the filename. ( #327 )
2016-11-11 22:39:05 -05:00
Davis King
e954a3a979
merged
2016-11-09 20:43:10 -05:00
Avi Haiat
85e2bcce02
Add option --rm-other-labels to imglab ( #323 )
...
* Add option --rm-other-labels to imglab
* Add support for comma separated list of labels
2016-11-09 20:36:52 -05:00
Davis King
8b87257103
Fixed rls numerical instability.
2016-11-09 10:30:16 -05:00
Davis King
b1909d5cc0
merged
2016-11-08 14:40:52 -05:00
Davis King
28d76d011f
Made rls run a bit faster, especially if the new mode that allows the
...
regularization to decay away is activated.
2016-11-08 14:40:19 -05:00
Davis King
c8c1abb733
Fixed matlab crashing when an error message from a mex file included the %
...
character, since that is interpreted by matlab as part of an eventual printf()
code.
2016-11-08 11:03:40 -05:00
Davis King
08a89c80c9
Make sure the state of the assert macros is always consistent between how dlib
...
is built as a stand alone library and what appears in the config.h file.
2016-11-07 19:50:12 -05:00
Davis King
1fe5aa0593
Adding more possible metric loss layers.
2016-11-06 13:53:22 -05:00
Davis King
1c19a80a39
Changed the loss layer interface to use two typedefs, output_label_type and
...
training_label_type instead of a single label_type. This way, the label
type used for training can be distinct from the type output by the network.
This change breaks backwards compatibility with the previous API.
2016-11-05 21:12:33 -04:00
Davis King
25ccbc4294
Added serialization support for std::array.
2016-11-05 10:52:09 -04:00
Davis King
cccde6324c
Changed CMake so it doesn't automatically enable asserts in debug builds since
...
this frustrates Visual Studio users who want to compile dlib as a static
library.
2016-11-04 20:30:03 -04:00
Davis King
bf94ce6f4e
merged
2016-11-04 20:16:30 -04:00
Davis King
3bb4a8e92d
merged
2016-11-04 20:16:08 -04:00
Stefan Schweter
02829ef903
Explicitly use signed char in max_cost_assignment test (fix for ARM/ARM64) ( #316 )
2016-11-04 16:55:25 -04:00
Stefan Schweter
a756183cee
Made active_learning test more robust on ARM64. ( #315 )
2016-11-04 13:48:09 -04:00
Stefan Schweter
a805051008
Update Travis CI status badge in README.md ( #311 )
2016-11-02 19:05:36 -04:00
Davis King
c1b389bca1
Fixed compile time error.
2016-11-02 10:12:07 -04:00
Davis King
45dd580bf4
Wrote replacements for set_tensor() and scale_tensor() since the previous
...
versions were calling into cuDNN, however, the cuDNN functions for doing this
are horrifically slow, well over 100x slower than they should be, which is
surprising since these functions are so trivial.
2016-11-02 10:06:36 -04:00
Davis King
44d67000fc
Made test deterministic
2016-10-30 07:42:27 -04:00
caseymcc
1cfcb4d836
fix include issue when using libjpeg compiled outside of dlib ( #307 )
2016-10-28 15:15:55 -04:00
Davis King
452b188def
Fixed some race conditions in cuda code.
2016-10-27 11:29:27 -04:00
Davis King
76d5bba613
Fixed compiler warnings
2016-10-26 17:09:44 -04:00
Davis King
3586d409f9
Added tt::scale_columns(). Also optimized some other cuda kernel launches a
...
little.
2016-10-26 17:06:02 -04:00
Davis King
2a2956a238
Added is_vector() for tensor objects.
2016-10-26 16:49:56 -04:00
Davis King
fa8b3767bf
Improved cudnn error messages.
2016-10-26 09:06:05 -04:00
Davis King
33af52d23a
updated docs
2016-10-23 15:43:47 -04:00
Davis King
b2faad0dd8
Added l2normalize_ layer.
2016-10-23 15:13:46 -04:00
Davis King
a390f10931
Made approximate derivative calculation slightly more numerically robust.
2016-10-23 10:44:05 -04:00
Davis King
ff9f597a60
Made the dnn_trainer automatically reload from the last good state if a loss of
...
NaN is encountered.
2016-10-21 22:48:03 -04:00
Davis King
4a2f0192e7
Fixed a bug in the repeat layer that caused it to throw exceptions
...
in some cases.
2016-10-18 22:54:34 -04:00
Davis King
85d421c356
Added --rmignore
2016-10-16 19:51:19 -04:00
Davis King
98bd454247
merged
2016-10-16 16:39:04 -04:00
Davis King
06cf15a0b5
Added loss_metric_
2016-10-16 16:38:17 -04:00
Davis King
0848616d99
Improved xml logging for --resample
2016-10-16 16:21:28 -04:00
Davis King
f8fcfcd0b4
removed --extract-chips
2016-10-16 16:19:24 -04:00
Davis King
968322bcf8
Fixed confusing -h docs
2016-10-16 16:10:44 -04:00
Davis King
ae88702ba6
Added --jpg and --one-object-per-image.
2016-10-16 16:09:06 -04:00
Evgeniy Fominov
97e3e420c8
Update Readme.md - travis badge ( #297 )
...
Added Travis badge 
2016-10-15 16:26:55 -04:00
Davis King
d695996416
enabling travis email notifications
2016-10-15 11:00:48 -04:00
Evgeniy Fominov
9bb271f170
Travis CI Integration scripts ( #294 )
2016-10-14 14:44:31 -04:00
hannometer
a8c51cea06
pass CUDA_HOST_COMPILER variable to cuda cmake tests ( #292 )
2016-10-14 08:35:09 -04:00
Mateusz Baran
41bc867a4d
optimization - poly_min_extrap fixed for a specific case ( #291 )
...
If temp is nonzero but a subnormal then temp2 matrix may contain infinities which may cause temp assignment in line 225 to be a NaN (as a difference of two positive infinities).
2016-10-14 06:54:59 -04:00
Davis King
d27511b40a
Record last changeset and set PATCH version to 99
2016-10-10 19:36:19 -04:00
Davis King
bfed73cc18
Added tag v19.2 for changeset f8fa027c7602
2016-10-10 19:34:33 -04:00
Davis King
5a970318c3
Created release v19.2
2016-10-10 19:34:32 -04:00
Davis King
ccdb77212a
Made tensor copies a little more efficient.
2016-10-10 18:26:09 -04:00
Davis King
a66a24f3e0
merged
2016-10-10 09:09:22 -04:00
Evgeniy Fominov
18fc97d324
Added threads.h into shape_predictor.h ( #283 )
2016-10-10 08:32:16 -04:00
