624 Commits (master)

Author	SHA1	Message	Date
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	5 years ago
Adrià Arrufat	57bb5eb58d	use running stats to track losses (#2041)	5 years ago
Davis King	0057461a62	Promote some of the sub-network methods into the add_loss_layer interface so users don't have to write .subnet() so often.	5 years ago
Adrià Arrufat	f42f100d0f	Add DCGAN example (#2035) ... * wip: dcgan-example * wip: dcgan-example * update example to use leaky_relu and remove bias from net * wip * it works! * add more comments * add visualization code * add example documentation * rename example * fix comment * better comment format * fix the noise generator seed * add message to hit enter for image generation * fix srand, too * add std::vector overload to update_parameters * improve training stability * better naming of variables make sure it is clear we update the generator with the discriminator's gradient using fake samples and true labels * fix comment: generator -> discriminator * update leaky_relu docs to match the relu ones * replace not with ! * add Davis' suggestions to make training more stable * use tensor instead of resizable_tensor * do not use dnn_trainer for discriminator	5 years ago
Adrià Arrufat	c832d3b2fc	simplify resnet definition by reusing struct template parameter (#2010) ... * simplify definition by reusing struct template parameter * put resnet into its own namespace * fix infer names * rename struct impl to def	5 years ago
Davis King	fc6992ac04	A little bit of cleanup	5 years ago
Adrià Arrufat	10d7f119ca	Add dnn_introduction3_ex (#1991) ... * Add dnn_introduction3_ex	5 years ago
Juha Reunanen	bd6994cc66	Add new loss layer for binary loss per pixel (#1976) ... * Add new loss layer for binary loss per pixel	5 years ago
Davis King	f2cd9e3b1d	use a time based exeuction limit in example	5 years ago
Juha Reunanen	d175c35074	Instance segmentation (#1918) ... * Add instance segmentation example - first version of training code * Add MMOD options; get rid of the cache approach, and instead load all MMOD rects upfront * Improve console output * Set filter count * Minor tweaking * Inference - first version, at least compiles! * Ignore overlapped boxes * Ignore even small instances * Set overlaps_ignore * Add TODO remarks * Revert "Set overlaps_ignore" This reverts commit 65adeff1f8. * Set result size * Set label image size * Take ignore-color into account * Fix the cropping rect's aspect ratio; also slightly expand the rect * Draw the largest findings last * Improve masking of the current instance * Add some perturbation to the inputs * Simplify ground-truth reading; fix random cropping * Read even class labels * Tweak default minibatch size * Learn only one class * Really train only instances of the selected class * Remove outdated TODO remark * Automatically skip images with no detections * Print to console what was found * Fix class index problem * Fix indentation * Allow to choose multiple classes * Draw rect in the color of the corresponding class * Write detector window classes to ostream; also group detection windows by class (when ostreaming) * Train a separate instance segmentation network for each classlabel * Use separate synchronization file for each seg net of each class * Allow more overlap * Fix sorting criterion * Fix interpolating the predicted mask * Improve bilinear interpolation: if output type is an integer, round instead of truncating * Add helpful comments * Ignore large aspect ratios; refactor the code; tweak some network parameters * Simplify the segmentation network structure; make the object detection network more complex in turn * Problem: CUDA errors not reported properly to console Solution: stop and join data loader threads even in case of exceptions * Minor parameters tweaking * Loss may have increased, even if prob_loss_increasing_thresh > prob_loss_increasing_thresh_max_value * Add previous_loss_values_dump_amount to previous_loss_values.size() when deciding if loss has been increasing * Improve behaviour when loss actually increased after disk sync * Revert some of the earlier change * Disregard dumped loss values only when deciding if learning rate should be shrunk, but *not* when deciding if loss has been going up since last disk sync * Revert "Revert some of the earlier change" This reverts commit 6c852124ef. * Keep enough previous loss values, until the disk sync * Fix maintaining the dumped (now "effectively disregarded") loss values count * Detect cats instead of aeroplanes * Add helpful logging * Clarify the intention and the code * Review fixes * Add operator== for the other pixel types as well; remove the inline * If available, use constexpr if * Revert "If available, use constexpr if" This reverts commit 503d4dd335. * Simplify code as per review comments * Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh * Clarify console output * Revert "Keep estimating steps_without_progress, even if steps_since_last_learning_rate_shrink < iter_without_progress_thresh" This reverts commit 9191ebc776. * To keep the changes to a bare minimum, revert the steps_since_last_learning_rate_shrink change after all (at least for now) * Even empty out some of the previous test loss values * Minor review fixes * Can't use C++14 features here * Do not use the struct name as a variable name	5 years ago
Davis King	1b83016abd	update docs	5 years ago
Davis King	39327e71b7	Added note about using cmake's new fetch content feature.	5 years ago
Davis King	fced3587f1	fixing grammar	5 years ago
Davis King	5d03b99a08	Changed to avoid compiler warning.	6 years ago
Juha Reunanen	f685cb4249	Add U-net style skip connections to the semantic-segmentation example (#1600) ... * Add concat_prev layer, and U-net example for semantic segmentation * Allow to supply mini-batch size as command-line parameter * Decrease default mini-batch size from 30 to 24 * Resize t1, if needed * Use DenseNet-style blocks instead of residual learning * Increase default mini-batch size to 50 * Increase default mini-batch size from 50 to 60 * Resize even during the backward step, if needed * Use resize_bilinear_gradient for the backward step * Fix function call ambiguity problem * Clear destination before adding gradient * Works OK-ish * Add more U-tags * Tweak default mini-batch size * Define a simpler network when using Microsoft Visual C++ compiler; clean up the DenseNet stuff (leaving it for a later PR) * Decrease default mini-batch size from 24 to 23 * Define separate dnn filename for MSVC++ and not * Add documentation for the resize_to_prev layer; move the implementation so that it comes after mult_prev * Fix previous typo * Minor formatting changes * Reverse the ordering of levels * Increase the learning-rate stopping criterion back to 1e-4 (was 1e-8) * Use more U-tags even on Windows * Minor formatting * Latest MSVC 2017 builds fast, so there's no need to limit the depth any longer * Tweak default mini-batch size again * Even though latest MSVC can now build the extra layers, it does not mean we should add them! * Fix naming	6 years ago
Juha Reunanen	cf5e25a95f	Problem: integer overflow when calculating sizes (may happen e.g. with very large images) (#1148) ... * Problem: integer overflow when calculating sizes (may happen e.g. with very large images) Solution: change some types from (unsigned) long to size_t # Conflicts: # dlib/dnn/tensor.h * Fix the fact that std::numeric_limits<unsigned long>::max() isn't always the same number * Revert serialization changes * Review fix: use long long instead of size_t * From long to long long all the way * Change more types to (hopefully) make the compiler happy * Change many more types to size_t * Change even more types to size_t * Minor type changes	7 years ago
Davis King	e6fe1e0259	merged	7 years ago
Davis King	c9faacce29	Fixed typos	7 years ago
Duc Thien Bui	9185e0a725	fix typo in train find cars example (#1028)	7 years ago
Davis King	efd945618d	Minor update	7 years ago
Davis King	603ebc2750	Changed this example to use repeat layers. This doesn't change the behavior of ... the code, but it helps visual studio use less RAM when building the example, and might make appveyor not crash. It's also a slightly cleaner way to write the code anyway.	7 years ago
Davis King	22f26ebe97	Improved visual studio compilation instructions	7 years ago
Davis King	46a1893534	Fixed spelling error in comment	7 years ago
visionworkz	ac292309c1	Exposed jitter_image in Python and added an example (#980) ... * Exposed jitter_image in Python and added an example * Return Numpy array directly * Require numpy during setup * Added install of Numpy before builds * Changed pip install for user only due to security issues. * Removed malloc * Made presence of Numpy during compile optional. * Conflict * Refactored get_face_chip/get_face_chips to use Numpy as well.	7 years ago
Davis King	e273f5159d	Added find_min_global() overloads.	7 years ago
Davis King	15c04ab224	This example still uses a lot of visual studio ram.	7 years ago
Davis King	2b3d8609e5	These examples compile now in visual studio due to the recent pragma directive added to core.h.	7 years ago
Davis King	4fa32903a6	clarified docs	7 years ago
Davis King	4d0b203541	Just moved the try block to reduce the indentation level.	7 years ago
Davis King	929870d3ad	Updated example to use C++11 style code and also to show the new find_max_global() routine.	7 years ago
Davis King	1aa6667481	Switched this example to use the svm C instead of nu trainer.	7 years ago
Davis King	04991b7da6	Made this example program use the new find_max_global() instead of grid search ... and BOBYQA. This greatly simplifies the example.	7 years ago
Amin Cheloh	1798e8877c	Update dnn_mmod_find_cars2_ex.cpp (#966)	7 years ago
Davis King	b84e2123d1	Changed network filename to something more descriptive.	7 years ago
Juha Reunanen	e48125c2a2	Add semantic segmentation example (#943) ... * Add example of semantic segmentation using the PASCAL VOC2012 dataset * Add note about Debug Information Format when using MSVC * Make the upsampling layers residual as well * Fix declaration order * Use a wider net * trainer.set_iterations_without_progress_threshold(5000); // (was 20000) * Add residual_up * Process entire directories of images (just easier to use) * Simplify network structure so that builds finish even on Visual Studio (faster, or at all) * Remove the training example from CMakeLists, because it's too much for the 32-bit MSVC++ compiler to handle * Remove the probably-now-unnecessary set_dnn_prefer_smallest_algorithms call * Review fix: remove the batch normalization layer from right before the loss * Review fix: point out that only the Visual C++ compiler has problems. Also expand the instructions how to run MSBuild.exe to circumvent the problems. * Review fix: use dlib::match_endings * Review fix: use dlib::join_rows. Also add some comments, and instructions where to download the pre-trained net from. * Review fix: make formatting comply with dlib style conventions. * Review fix: output training parameters. * Review fix: remove #ifndef __INTELLISENSE__ * Review fix: use std::string instead of char* * Review fix: update interpolation_abstract.h to say that extract_image_chips can now take the interpolation method as a parameter * Fix whitespace formatting * Add more comments * Fix finding image files for inference * Resize inference test output to the size of the input; add clarifying remarks * Resize net output even in calculate_accuracy * After all crop the net output instead of resizing it by interpolation * For clarity, add an empty line in the console output	7 years ago
Davis King	978da26ed0	Fixed grammar in comment	7 years ago
Davis King	50de3da992	Updated comments to reflect recent API changes.	7 years ago
Davis King	dc0245af05	Changed graph construction for chinese_whispers() so that each face is always ... included in the edge graph. If it isn't then the output labels from chinese_whispers would be missing faces in this degenerate case.	7 years ago
Davis King	e338bf02e0	Changed the random_cropper's set_min_object_size() routine to take min box ... dimensions in the same format as the mmod_options object (i.e. two lengths measured in pixels). This should make defining random_cropping strategies that are consistent with MMOD settings much more straightforward since you can just take the mmod_options settings and give them to the random_cropper and it will do the right thing.	7 years ago
Davis King	1c664eeac5	Made the metric learning example do image jittering.	7 years ago
Davis King	369f2b32e8	Cleaned up jitter_image() code and moved it into dlib proper.	7 years ago
Davis King	525bf0e15b	Fixed weird wording.	7 years ago
Davis King	3aa9991892	Made comment more clear.	7 years ago
Davis King	3b2d73db68	Some cmake script cleanup and refactoring.	7 years ago
Davis King	d39b843de0	Added another car finding example	7 years ago
Davis King	4737009c11	Clarified comments	7 years ago
Davis King	8bfd227930	Changed to slightly better default recommendations for parameters.	7 years ago
Davis King	b4bd6f8dee	More cleanup	7 years ago
Davis King	57b6cd1b80	clarified error message	7 years ago
Davis King	37e0b9ea68	Updated comments.	7 years ago
Davis King	41c2f217c6	Changed code to recommend users use the new 5 point face landmarking model.	7 years ago
Davis King	4fcc5faa39	Fixed grammar	7 years ago
Davis King	8de1a1ed6a	Improved citations	7 years ago
Davis King	71db541a92	Minor cleanup	7 years ago
Davis King	3211da440d	Yet more comments	7 years ago
Davis King	a362305e1b	cleanup	7 years ago
Davis King	5e9f094d5a	Fixed compiler warning	7 years ago
Davis King	88c1bf3b46	Added youtube links	7 years ago
Davis King	efb1d83dd3	More comments	7 years ago
Davis King	54de7a8546	Added more comments	7 years ago
Davis King	aec695fc26	Cleanup	7 years ago
Davis King	51eae2ba07	Added two vehicle detection examples.	7 years ago
Davis King	420eba0e6a	Added note about logging training parameters.	7 years ago
Davis King	b4853ce449	Fixed grammar	7 years ago
MambaWong	373ca59369	Update webcam_face_pose_ex.cpp (#702) ... Test on a given video like this cv::VideoCapture cap("Sample.avi") may be broken when the video frames are not enough before the main window is closed by the user.	7 years ago
Davis King	64052a3ff1	Added a comment	7 years ago
Davis King	f5574434db	Upgraded loss_mmod_ to support objects of varying aspect ratio. This changes ... the API for the mmod_options struct slightly.	7 years ago
Davis King	17b48b97bb	Changed the random_cropper's interface so that instead of talking in terms of ... min and max object height, it's now min and max object size. This way, if you have objects that are short and wide (i.e. objects where the relevant dimension is width rather than height) you will get sensible behavior out of the random cropper.	7 years ago
Davis King	334ba38ec6	A minor change to hopefully reduce inane warnings from visual studio.	7 years ago
elelel	b57b8b20aa	Migrate from dlib::scoped_ptr to std::unique_ptr (#593) ... * Convert unique_ptr * Fix passing unique_ptr as copy by value * Remove scoped_ptr implementation * Fix missed files * Move bool cast into tester macro * Reexport scoped_ptr alias from sockets	7 years ago
elelel	74f9f2fb2a	Replace shared_ptr/weak_ptr with STL's versions (#588) ... * Replace shared_ptr/weak_ptr with stdlib counterparts * Fix ptr usage through tests compilation * Bring back dlib smart ptrs as legacy * Include scoped_ptr directly * Add explanation about smart_ptr deprecation	7 years ago
Davis King	70b885eb0b	Added a note about exporting to XML.	7 years ago
Davis King	482c2db7ba	Changed the location of the set_synchronization_file() call so that rerunning ... the example restarts correctly.	7 years ago
Davis King	a1eea964b6	Added more comments	7 years ago
cclough	5fa4533b9a	Update CMakeLists.txt (#566)	7 years ago
Davis King	df5a087b03	Made it more obvious that this file is a cmake tutorial. Also added a few more comments about how to use cmake.	8 years ago
Davis King	67db08f5a6	Made example cmake use dlib::dlib link name. Also made the installed ... cmake config for dlib enable C++11.	8 years ago
Alain Vaucher	b41455fd26	Change minimum CMake version that supports the introduced features	8 years ago
Davis King	e4402cf575	Don't build face recognition example if GUI support isn't enabled.	8 years ago
Davis King	7747be8302	cleanup	8 years ago
Davis King	11016884ed	Avoid crashing visual studio	8 years ago
Davis King	c5b29d985b	Added try/catch block to main	8 years ago
Davis King	d7a6cd4960	fixed grammar in comment	8 years ago
Davis King	fe1a15f32a	Added face recognition example	8 years ago
Davis King	45d047de80	Minor cleanup	8 years ago
Davis King	57ca3e5459	Made network smaller.	8 years ago
Davis King	6c6c25a344	merged	8 years ago
RangerUFO	0f60ba365b	Fix comments typo in dnn_introduction2_ex (#405)	8 years ago
Davis King	3793e29e0e	Added comments about test_one_step()	8 years ago
Davis King	bf36a846de	Changed build order to see if it prevents travis from crashing.	8 years ago
Davis King	093f709a9a	Minor improvement to example	8 years ago
Davis King	ab1575da20	more cleanup	8 years ago
Davis King	b87ecad51e	Improved example	8 years ago
Davis King	fd13230486	Finished the more complex metric learning example and added some example data.	8 years ago
Davis King	f4b3c7ee0f	Improved example	8 years ago
Davis King	f28d2f7329	Cleaned up example	8 years ago
Davis King	4d121e7e08	Added another metric learning example	8 years ago
Davis King	cb198afc3f	Added metric learning example	8 years ago
Davis King	2787edc9bd	Improved error messages about non-C++11 compilers and made the example cmake ... file a little clearer on this issue.	8 years ago
Davis King	f5da9fe2b0	Changed cmake so it compiles most of the DNN examples when using visual studio.	8 years ago