7863 Commits (e7c25c06df676a19e841897d3db7779aa24f4d88)

 

All Branches   

Search

Author	SHA1	Message	Date
Sajied Shah Yousuf	e7c25c06df	Changed directory of license (#2189) ... Add copy of license file to root to make github happy.	4 years ago
Davis King	20a1477209	update docs	4 years ago
pfeatherstone	ab346ddfa6	Extended proxy_(de)serialize objects to work with stringstream, ostringstream, istringstream and vector<char> (#2181) ... * [DLIB] extended proxy objects to work with strinstream, istringstream, ostringstream and vector<char> * [DLIB] - use std::istream and std::ostream instead of std::istringstream, std::ostringstream and std::stringstream. - put back the filename member variable for better error messages * [DLIB] - review requirement Co-authored-by: pf <pf@pf-ubuntu-dev>	4 years ago
Adrià Arrufat	fa818b9a96	use DLIB_CASSERT to avoid unused variable warning in release compilation (#2182)	4 years ago
pfeatherstone	d4fe74b5a8	vectorstream updates: added seekoff and seekpos (#2179) ... * [DLIB] added seekpos and seekoff functions. These are necessary for functions in iostream base class to work properly. e.g. seekg. Note that in seekoff, you do NOT want to check the validity of read_pos after it has been updated. dlib::vectorstream and std::iostream work together to set EOF and/or badbit. Doing something like seekg(10000) should not throw even if the underlying buffer has 2 bytes. You should check if EOF is set and possibly call clear(). We have removed seekg from dlib::vectorstream as this adds confusion. Now std::iostream::seekg is called which somewhere down the callstack will call seekpos and/or seekoff. So there should be no diverging functionality between calling seekg on dlib::vectorstream& or std::iostream& when there is a cast. * [DLIB] vectorstream unit test is updated to run identical tests on dlib::vectorstream& and std::iostream& * [DLIB] only support read pointers and delete copy and move semantics * [DLIB] explicit tests for seekg() in different directions * [DLIB] - no need to delete the move constructor and move assign operator. This is implicitly done by deleting the copy constructor and copy assign operator. * [DLIB] - remove leftover comments. no need - use more idiomatic notation Co-authored-by: pf <pf@pf-ubuntu-dev>	4 years ago
Davis King	cdeb2e067c	add some docs	4 years ago
pfeatherstone	12a82f6542	Macro for generating default serialisation functions (#2177) ... * [DLIB] macro for generating default serialisation functions * [DLIB] refactoring * [DLIB] refactoring	4 years ago
Adrià Arrufat	9d60949a3a	Add scale_prev layer (#2171) ... * Add scale_prev layer * remove comment and fix gradient * add test for scale_ and scale_prev_ layers	4 years ago
Adrià Arrufat	77e6255fdd	Add error message for mismatched tensor sizes in dnn_trainer (#2165)	4 years ago
Davis King	40c3e48818	Simplified more uses of layer visiting and fixed constness bug ... The const bug was introduced yesterday and caused some layer visiting to not work on const networks.	4 years ago
Adrià Arrufat	5ec60a91c4	Show how to use the new visitors with lambdas (#2162)	4 years ago
Davis King	393db2490b	switch this to C++11 code	4 years ago
Davis King	5bcbe617eb	make type_safe_union movable and also support holding movable types in a natural way.	4 years ago
Davis King	afe19fcb8b	Made the DNN layer visiting routines more convenient. ... Now the user doesn't have to supply a visitor capable of visiting all layers, but instead just the ones they are interested in. Also added visit_computational_layers() and visit_computational_layers_range() since those capture a very common use case more concisely than visit_layers(). That is, users generally want to mess with the computational layers specifically as those are the stateful layers.	4 years ago
Davis King	7dcc7b4ebc	Added call_if_valid()	4 years ago
Adrià Arrufat	e7ec6b7777	Add visitor to remove bias from bn_ layer inputs (#closes 2155) (#2156) ... * add visitor to remove bias from bn_ inputs (#closes 2155) * remove unused parameter and make documentation more clear * remove bias from bn_ layers too and use better name * let the batch norm keep their bias, use even better name * be more consistent with impl naming * remove default constructor * do not use method to prevent some errors * add disable bias method to pertinent layers * update dcgan example - grammar - print number of network parameters to be able to check bias is not allocated - at the end, give feedback to the user about what the discriminator thinks about each generated sample * fix fc_ logic * add documentation * add bias_is_disabled methods and update to_xml * print use_bias=false when bias is disabled	4 years ago
Davis King	ed22f0400a	Make dnn_trainer use robust statistic to determine if the loss is exploding and if it should backtrack. ... Previously we used only the non-robust version, and so would mistakenly not catch sequenes of loss increase that begin with an extremely large value and then settled down to still large but less extreme values.	4 years ago
Davis King	0bb6ce36d8	dnn_trainer prints number of steps executed when print to ostream	4 years ago
Davis King	76cc8e3b6b	Add probability_values_are_increasing() and probability_values_are_increasing_robust()	4 years ago
Davis King	c14ba4847e	Rename POSIX macro to DLIB_POSIX to avoid name clashes with some libraries.	4 years ago
Davis King	4b92804dc2	Use the box with bounding box regression applied to do NMS in the loss.	4 years ago
Davis King	0e721e5cae	Fix bug in bounding box regression loss.	4 years ago
Adrià Arrufat	c9809e067f	Add missing input/output mappings to mult_prev (#2154)	4 years ago
Davis King	b401185aa5	Fix a warning and add some more error handling.	4 years ago
Adrià Arrufat	dd06c1169b	loss multibinary log (#2141) ... * add loss_multilabel_log * add alias template for loss_multilabel_log * add missing assert * increment truth iterator * rename loss to loss_multibinary_log * rename loss to loss_multibinary_log * explicitly capture dims in lambda	4 years ago
Juha Reunanen	d7ca478b79	Problem: With certain batch size / device count combinations, batches were generated with size = 1, causing problems when using batch normalization. (#2152) ... Solution: Divide the mini-batch more uniformly across the different devices.	4 years ago
Davis King	bea99ceed0	switch to a name less likely to conflict with third party code	4 years ago
Juha Reunanen	a9592b07fd	Minor typo fixes (#2150)	4 years ago
samaldana	2a870e329c	Fix warning for zero variadic macro arguments. (#2151) ... When consuming dlib headers and building using gcc/clang with flags '-Werror -Wpedantic', any inclusion involving DLIB_CASSERT triggers a compilation error: ISO C++11 requires at least one argument for the "..." in a variadic macro Co-authored-by: Samuel Aldana <samuel.aldana@cognex.com>	4 years ago
pfeatherstone	f3b4fc548d	Added "get_random_complex_gaussian" to dlib::rand (#2149) ... * Added a function for computing a gaussian distributed complex number. The real version is adapted to use the complex version * Missing header * missed std:: I was too quick Co-authored-by: pf <pf@pf-ubuntu-dev>	4 years ago
Davis King	f55a1a51a0	fix python code index page. ... The recent change to use a dlib/__init__.py file instead of the dlib.so file directly messed it up.	4 years ago
Davis King	59b44849bd	fix typo, doesn't really matter, but still	4 years ago
Davis King	02e70ce3ca	Record last changeset and set PATCH version to 99	4 years ago
Davis King	9117bd7843	Created release v19.21	4 years ago
Davis King	2e64bdd449	update docs	4 years ago
Davis King	2c70aad12c	Use a cache to avoid calls to the cuDNN algorithm selection routines.	4 years ago
Davis King	8910445a7a	fix some spelling and grammar errors	4 years ago
Davis King	4721075314	More optimization unit tests	4 years ago
Davis King	a9d554a4ac	minor cleanup	4 years ago
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.	4 years ago
Davis King	c90362d852	updated release notes	4 years ago
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.	4 years ago
Davis King	f8cfe63904	Avoid unnecessairly asking cuDNN which algorithms to use, since this is slow in cuDNN 8.0	4 years ago
Davis King	6c3243f766	Cleanup cuDNN conv algorithm selection code slightly by moving it into its own function.	4 years ago
Davis King	4d18e0d0c7	oops, fixing a weird typo	4 years ago
Davis King	3400e163e8	tweaked cca test thresholds to avoid false positives	4 years ago
Davis King	943408d2d2	Allow forwarding initial function evaluations into find_max_global()	4 years ago
Davis King	5a80ca9e5f	Apply --expt-relaxed-constexpr to all older versions of cuda.	4 years ago
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)	4 years ago
Davis King	23b9abd07a	Switch cuda target architecture from sm_30 to sm_50. I.e. Maxwell instead of Kepler.	4 years ago