Commit Graph

5970 Commits

Author SHA1 Message Date
Davis King
0c0bde22a8 New versions of boost and cmake need to be told to look in boost's library folder to find boost's libraries
when building on windows.  Bizarre.
2016-08-13 12:47:55 -04:00
Davis King
02b209316a Added workaround for a bug in cuDNN5.1 which causes
cudnnGetConvolutionBackwardFilterAlgorithm() to pick invalid algorithms,
resulting in cuDNN not working correctly.
2016-08-11 22:12:40 -04:00
Davis King
fde662b31e updated docs 2016-08-11 19:07:30 -04:00
Davis King
480307d03f Added additional search names for boost-python 2016-08-10 09:30:04 -04:00
Davis King
ceb6a119d5 Fixed a bug in the cuDNN binding that on rare occasions caused NaN outputs from
batch normalization.  The running mean and variance need to be initialized when
calling cuDNN, even if the averaging factor is 1.  I don't think this was the
case prior to cuDNN v5, but it certainly is the case now.  This patch fixes
this.
2016-08-09 16:33:44 -04:00
Davis King
98da589f9d merged 2016-08-09 10:48:48 -04:00
Davis King
e6a7f14c0e made unit test more repeatable 2016-08-09 10:36:38 -04:00
Davis King
f01f02b2a5 Added more testing messages 2016-08-06 12:39:13 -04:00
Davis King
2dd89e3f04 CMake change to enable cuda use on ubuntu 16.06 2016-08-06 12:00:41 -04:00
Davis King
e0af06c0db added missing #include file 2016-08-05 19:14:49 -04:00
Davis King
36a0776ad3 Merge branch 'sutr90-lab_pixel' 2016-08-05 08:14:20 -04:00
Davis King
8332489805 Fixed a few places where lab color space still referred to the channels
as being signed rather than unsigned.
2016-08-05 08:13:45 -04:00
Davis King
8db76eb89a Merge branch 'lab_pixel' of git://github.com/sutr90/dlib into sutr90-lab_pixel 2016-08-05 07:11:20 -04:00
sutr90
64e437a320 Added test for RGB to Lab to RGB conversion. 2016-08-05 10:59:07 +02:00
Davis King
8c037c3cb0 Added guards so dlib::async code isn't pulled in when using old versions of visual studio (since they don't support c++11). 2016-08-04 21:12:00 -04:00
Davis King
a5e264cb31 Added check for initializer list support so code compiles in visual studio. 2016-08-04 21:10:12 -04:00
Davis King
87e04764e5 Removed call to std::round() since visual studio 2012 doesn't support it. 2016-08-04 21:09:52 -04:00
sutr90
fa865c218e Added tests for Lab pixels. 2016-08-03 14:48:59 +02:00
Davis King
bc4b2814af Fixed silly typo in script 2016-07-31 09:49:21 -04:00
Davis King
f2088e832b removed excessive error checking 2016-07-31 09:46:10 -04:00
Davis King
6402a37458 Made --resample use a 64bit hash of the image data
in the file names to make it so there won't ever be any name conflicts
in the cropped images.
2016-07-31 09:11:52 -04:00
Davis King
7c828d0812 Added some scripts that help with imglab XML file management. 2016-07-30 18:33:50 -04:00
Davis King
f77ca0c754 Added --files 2016-07-30 17:19:20 -04:00
Davis King
26c727c79e Added --ignore option. Also improved how scale jittering happens in --resample,
now the output images are always the same dimensions but the objects scale changes.
2016-07-30 14:34:53 -04:00
Davis King
ed78e8b1c0 Made this not use stdin's file descriptor for data transfer between the
processes since sometimes stdin was closed in some environments.  Also cleaned
up the code a little bit.
2016-07-29 09:03:05 -04:00
Davis King
d40e34cf22 merged 2016-07-28 19:13:10 -04:00
Davis King
284ace058a merged 2016-07-28 19:12:48 -04:00
Davis King
3a096469ef Changed code to avoid advancing iterator beyond end since some compilers
complain about this (and it's technically not allowed in C++).
2016-07-28 19:09:07 -04:00
Davis King
edb2790bcc fixed grammar in comment 2016-07-28 17:59:31 -04:00
sutr90
933ab00942 Fixed PNG saving. 2016-07-28 14:07:23 +02:00
sutr90
2c6351bea3 Fixes Lab tests. 2016-07-27 17:51:01 +02:00
sutr90
ebad74aa20 Fixed grayscale assignment. 2016-07-27 17:47:42 +02:00
sutr90
6e06fc873c Changed lab pixel range.
Added clamping on conversion functions.
2016-07-27 17:25:19 +02:00
sutr90
71ab04f7bc Added missing conversion between HSI and CIELab. 2016-07-26 22:11:56 +02:00
sutr90
385bcb7660 Added support for CIELab color space pixels. 2016-07-26 21:45:52 +02:00
Jan Rüegg
d83b54d275 Make filo compile with WIN32_LEAN_AND_MEAN defined (#169) 2016-07-26 08:30:20 -04:00
Davis King
435299081f Fixed potential division by zero. 2016-07-23 20:18:31 -04:00
Davis King
938e97713a fixed typo in comment 2016-07-23 13:34:08 -04:00
Davis King
0f06d78e33 Add guard so code compiles in old compilers. 2016-07-22 23:01:33 -04:00
Davis King
d635a7ee10 Improved cmake warning messages about cuDNN 2016-07-22 17:03:34 -04:00
Davis King
73156146f9 Added dlib::async() and default_thread_pool() 2016-07-22 16:24:25 -04:00
Davis King
bdbc8e418b Renamed something to avoid name clash with standard library. 2016-07-22 16:22:57 -04:00
Evgeniy Fominov
bbeac285d1 Shape predictor trainer optimizations (#126)
* Shape predictor trainer optimizations

* Fixed performance leak in single thread mode & made VS2010 support
2016-07-22 09:11:13 -04:00
Davis King
6eb5bd8068 merged 2016-07-21 22:33:43 -04:00
Davis King
5ecd904d76 merged 2016-07-21 22:33:27 -04:00
Davis King
cf321559e8 Made remove_unobtainable_rectangles() work with all scan_fhog_pyramid objects. 2016-07-21 22:32:48 -04:00
Davis King
5f48afc5b8 Improved --extract-chips and --resample a bit. 2016-07-21 22:32:25 -04:00
Davis King
fd6db8216a Made it so the matrix only uses matlab's allocation routines for matrices that
are directly the arguments of a mex function.  This way, we avoid the problem
of dlib::matrix objects being created in other threads that internally call
into matlab and mess it up (since matlab is not thread safe in any sense).
2016-07-21 17:10:51 -04:00
Davis King
41d261acb6 merged 2016-07-20 20:29:18 -04:00
Davis King
d0fae99107 merged 2016-07-20 20:28:47 -04:00