OrgLion-ML
/
dlib
mirror of https://github.com/davisking/dlib.git
1
0
Fork 0
Code Issues Releases Wiki Activity
8,031 Commits 4 Branches 61 Tags 140 MiB
Commit Graph

8031 Commits

Author SHA1 Message Date
Davis King 02e70ce3ca Record last changeset and set PATCH version to 99 2020-08-08 15:30:37 -04:00
Davis King 9117bd7843 Created release v19.21 2020-08-08 15:26:07 -04:00
Davis King 2e64bdd449 update docs 2020-08-08 15:25:53 -04:00
Davis King 2c70aad12c Use a cache to avoid calls to the cuDNN algorithm selection routines. 2020-08-07 16:24:28 -04:00
Davis King 8910445a7a fix some spelling and grammar errors 2020-08-07 15:41:42 -04:00
Davis King 4721075314 More optimization unit tests 2020-08-07 09:57:12 -04:00
Davis King a9d554a4ac minor cleanup 2020-08-05 08:13:58 -04:00
yuriio ff3023f266
Added possibility to load PNG images from a data buffer. (#2137) 
* Added possibility to load PNG images from a data buffer.

* Fixed code not compiling with some versions of libpng that doesn't have const specifier.

* Used FileInfo struct as a single parameter for the read_image method.
 2020-08-05 08:11:46 -04:00
Davis King c90362d852 updated release notes 2020-08-02 08:20:44 -04:00
Davis King 7b564927d6 Switching to what is hopefully a better fix for the following CUDA error 
error: calling a constexpr host function("log1p") from a device function("cuda_log1pexp") is not allowed. The experimental flag '--expt-relaxed-constexpr' can be used to allow this.

The error only happens with some versions of CUDA.
 2020-08-01 13:48:30 -04:00
Davis King f8cfe63904 Avoid unnecessairly asking cuDNN which algorithms to use, since this is slow in cuDNN 8.0 2020-08-01 13:45:38 -04:00
Davis King 6c3243f766 Cleanup cuDNN conv algorithm selection code slightly by moving it into its own function. 2020-08-01 13:33:39 -04:00
Davis King 4d18e0d0c7 oops, fixing a weird typo 2020-07-26 15:13:20 -04:00
Davis King 3400e163e8 tweaked cca test thresholds to avoid false positives 2020-07-26 12:43:21 -04:00
Davis King 943408d2d2 Allow forwarding initial function evaluations into find_max_global() 2020-07-26 12:43:21 -04:00
Davis King 5a80ca9e5f Apply --expt-relaxed-constexpr to all older versions of cuda. 2020-07-24 23:50:22 -04:00
jbfove 5650ce45a1
Fix restoration of MSVC warnings in public headers (#2135) 
Previously they were restored to default values, which had the effect of negating the current setting of the calling code (whether set in the compiler options or by pragma previously)
 2020-07-22 06:07:49 -04:00
Davis King 23b9abd07a Switch cuda target architecture from sm_30 to sm_50. I.e. Maxwell instead of Kepler. 2020-07-11 21:07:36 -04:00
stoperro a2498dc47c
Additional documentation for failed dlib::layer<> use. (#2118) 2020-06-28 11:35:15 -04:00
Davis King 71b303d883 Improve the error a user gets when trying to build the pyhton extention on windows but they didn't install visual C++. 2020-06-27 20:08:39 -04:00
Davis King b9f4da5522 Make cuDNN test project failure print a message saying exactly why it failed. 2020-06-21 08:48:02 -04:00
Davis King facefa0204 Fix random foreset regression not doing quite the right thing. 2020-06-20 14:44:30 -04:00
Davis King fe803b566f add support for cudnn 8.0 2020-06-20 09:43:17 -04:00
Davis King f8887d8cbb fix paths 2020-06-17 21:29:10 -04:00
Davis King 2a3cb83afc Make dlib/__init__.py know the paths to cuda that cmake used. Use those 
paths instead of relying on CUDA_PATH to be set.
 2020-06-17 20:17:10 -04:00
Davis King 1515adc744 work around a bug in nvcc 2020-06-10 08:19:59 -04:00
Davis King 7fe355e0ec Oops, put import statements after path adjustment 2020-06-07 16:59:40 -04:00
Davis King 7fcb6c285d Add an __init__.py file so we can add paths to CUDA libs on windows. 2020-06-07 16:42:44 -04:00
Davis King 53b6ea3bf5 Record last changeset and set PATCH version to 99 2020-06-06 14:58:34 -04:00
Davis King 5612caa169 Created release v19.20 2020-06-06 14:53:52 -04:00
Davis King 610aa6349a updated docs 2020-06-06 14:52:21 -04:00
Davis King 3d738e1c2d sanitize git logs for export to html 2020-06-06 14:26:27 -04:00
Davis King 883101477d minor cleanup 2020-06-01 08:36:16 -04:00
stoperro a83242014e
Corrected interpolate_bilinear for lab_pixel. (#2091) 
* * Corrected interpolate_bilinear for non-RGB images not to collapse into grayscale (#2089)

* * interpolate_bilinear uses now pixel_to_vector for shorter code.

* pixels now have operator!=.

* * Explicitely float interpolation

* Using C++11 static_assert() in interpolation.

* * Corrected documentation for interpolate_bilinear, interpolate_quadratic

* * Corrected formatting near interpolate_bilinear
 2020-06-01 08:35:44 -04:00
Davis King 693aa0a719 fix build errors in cuda 10.2 2020-05-14 22:06:18 -04:00
Juha Reunanen c7062aa363
Minor optimization: add shortcut to in-place image resize if size_scale is 1 (#2076) 2020-05-04 21:10:31 -04:00
Adrià Arrufat 64ba66e1c7
fix receptive field comment (#2070) 2020-04-27 06:02:26 -04:00
ncoder-1 8055b8d19a
Update dnn_introduction_ex.cpp (#2066) 
Changed C-style cast to static_cast.
 2020-04-22 07:37:58 -04:00
Davis King 253745d29f fix typo in comment 2020-04-19 13:57:16 -04:00
Davis King a2e45f00b2 Reduce code duplication a bit and make equal_error_rate() give correct results when called on data where all detection scores are identical. 
Previously it would say the EER was 0, but really it should have said 1 in this case.
 2020-04-18 13:57:56 -04:00
Davis King 0e923cff93 A little bit of cleanup 2020-04-18 09:30:59 -04:00
Adrià Arrufat 55e9c890fd
Add cuda implementation for loss_mean_squared_per_channel_and_pixel (#2053) 
* wip: attempt to use cuda for loss mse channel

* wip: maybe this is a step in the right direction

* Try to fix dereferencing the truth data (#1)

* Try to fix dereferencing the truth data

* Fix memory layout

* fix loss scaling and update tests

* rename temp1 to temp

* readd lambda captures for output_width and output_height

clangd was complaining about this, and suggested me to remove them
in the first, place:

```
Lambda capture 'output_height' is not required to be captured for this use (fix available)
Lambda capture 'output_width' is not required to be captured for this use (fix available)
```

* add a weighted_loss typedef to loss_multiclass_log_weighted_ for consistency

* update docs for weighted losses

* refactor multi channel loss and add cpu-cuda tests

* make operator() const

* make error relative to the loss value

Co-authored-by: Juha Reunanen <juha.reunanen@tomaattinen.com>
 2020-04-18 09:29:46 -04:00
Davis King b42722a75d Fix DLIB_ISO_CPP_ONLY not working 2020-04-14 07:49:28 -04:00
Davis King fbb2db2188 fix example cmake script 2020-04-04 09:55:08 -04:00
Adrià Arrufat b44d9465f6
Fix warning in dnn_trainer initialization list (#2049) 
The thread pool was initialized after the network, so it lead to a
reorder warning in GCC 9.3.0
 2020-04-03 07:58:05 -04:00
Adrià Arrufat 5a715fe24d
Remove outdated comment from DCGAN example (#2048) 
* Remove outdated comment

That comment was there from when I was using a dnn_trainer to train
the discriminator network.

* Fix case
 2020-04-02 07:14:42 -04:00
Davis King 237746fc13 disable in source builds 2020-03-31 19:41:38 -04:00
Adrià Arrufat e9c56fb21a
Fix warnings while running the tests (#2046) 
* fix some warnings when running tests

* rever changes in CMakeLists.txt

* update example make use of newly promoted method

* update tests to make use of newly promoted methods
 2020-03-31 19:35:23 -04:00
Adrià Arrufat d1d96e380c
remove branch from cuda kernel (#2045) 
* remove branch from cuda kernel

* promote lambda to a global function
 2020-03-31 19:33:25 -04:00
Adrià Arrufat 57bb5eb58d
use running stats to track losses (#2041) 2020-03-30 20:20:50 -04:00