Mitie.py in python 2.7 binary does not match Mitie.py master #154
Comments
|
Yeah, it's been a while since I uploaded new precompiled binaries. It's probably about time to refresh them :) |
|
Can you please share the link of uploaded binaries. I am not seeing any updates in the download section |
|
I haven't updated anything yet. Be patient. |
peter277
added a commit
to peter277/MITIE
that referenced
this issue
Mar 28, 2025
636c0bcd1 Change version
d3520131a change the default u32string char type to char32_t (#3059)
8dab9c385 Check image size when reading targa file (#3058)
e8b6c05b9 Slice layer (#3055)
ef3d636af Update minimum CMake version to 3.10.0 (#3052)
28c46fb85 minor tweaks to the transformer example (#3048)
8fdd2a677 Add a first example for the new Dlib layers to build a transform-type network (#3041)
d6706a5f5 find blas and cudnn update for windows OS (#3047)
8454765a7 Replace deprecated pkgutil.find_loader() (#3043)
daefd41b9 Adjust github ci runner images (#3040)
230c0b019 Add multm_prev_ layer and enhance gemm() function for PLANE_WISE operations (#3020)
dfbee6d3c Fix saving JXL generic grayscale images (#3032)
39240959f Make event_correlation() work on fractional counts
6d29e0c7d Remove vscode stuff (#3028)
856707536 Add embeddings_ layer and supporting utility functions (#3021)
488ee5c59 handle empty proto objects
44689c1e9 added docs
fa5d37a62 add serialization support to chip_details (#3026)
4e53f8316 Add tril_ layer for lower triangular matrix operations (#3018)
72822fe4f Fix stride indexing bugs in `reorg` and `reorg_gradient` functions (CPU & CUDA) (#3012)
90c8d7830 Add Transpose Layer for 2D matrix transposition in 4D tensors (#3013)
fe6e052e0 Cleanup: remove using namespace std; (#3016)
fafdac37f Add RMS Normalization Layer (#2999)
253098eb1 Fix layer_normalize gradients (#3001)
27a013522 Add customizable dropout layer with compile-time rate specification (#3000)
98d26dfaf change the version back
986889d97 make a new dlib version to get us cmake installer instructions for python users
aba96b576 Give python users clear instructions about how to install cmake
3ca155da6 [FFMPEG] support ffmpeg 7 (#2987)
58b50552f put version back to .99
a2eec6aa1 just to tag v19.24.5
c2364600e remove deprecated setuptools test stuff
6ae301a11 Disable FFmpeg support when building the Python wheel (#2978)
0947215dd global_optimization: add template argument list (#2973)
6233808ce Update Pybind11 to v2.12.0 for numpy v2 support (#2970)
1a51998ff Make docs more clear
51c7a3597 Add input_tensor input type (#2951)
fa0e3ff95 Hide windows.h in .cc file (#2952)
e9bd9b864 Add support for riscv64 (#2949)
87da2959d suppress another one
b9355f08c suppress incorrect warning
48e633170 fixed bug in decoder (#2946)
2dbc7b9f7 fix typo (which turns out builds on gcc)
73c990dd1 Use size_t to avoid possible numeric overflow for the very largest images.
792b74440 better docs
ca7fd9d1a Actually load rgb_alpha images. (#2942)
8edfa542b Remove MMX header (#2939)
f8bf6caef Fix incorrect version check for libgif (#2936)
0af9e7a8d adjust head back to .99
de5c1c866 tagging another minor version
1db5a3853 Reduce FOREVER from 290 to 200 years (#2938)
4d6b41c7a Update HTML docs more
14ba5572e CMake: Set LAPACK libs as public (#2923)
c26b1238d jxl: use float for quality (#2930)
1529b690d Apply the new patch from Davis (#2929)
d5909ed97 allow reading and writing rgb-alpha images in Python (#2925)
d45f534d6 Add missing magic number for JPEG XL (#2924)
9a30c6d48 Add JXL support (#2917)
c7ba757bc add nr and nc methods to jpeg, png and webp loaders (#2920)
405536051 Use STD in testing suite (#2922)
3d3bd7cee Fix documentation of inv_color_transform (#2918)
f775b55cd Add color transform and inv color transform (#2914)
631c47c7f fix letterbox_image in yolo example (#2911)
85aa29a4e Return shortcut in disturb_colors (#2909)
b0f6be805 loss_barlow_twins: add get_eccm member function (#2906)
46e59a217 Disable mac osx CI
19a952c3a extend letterbox behavior (#2899)
b20e97446 Add find_convex_hull using Graham scan (#2889)
f7d99ae0d Make use of CMAKE_BUILD_PARALLEL_LEVEL (#2879)
13c6b7a5b Remove cmake from pyproject.toml (#2878)
106c5a265 [PNG] save_png() now works with std::vector and std::ostream (#2873)
3624bf9f0 [scope] Adding scope_exit (#2875)
afede571c potential fix for gcc 5.1 onward (#2870)
57b77a0e4 fix incorrect assert and typo in docs
b74a6710d simplify and fix bug (#2866)
dad29bfae Apple Clang complains about sprintf. Use snprintf instead. (#2862)
c53ff2595 bug fix when multiple streams are present (#2863)
4161cc301 oops, fix build error
1ffab7c21 update docs and some cleanup
b4acae25a Apply archiver flags only on Apple using Apple compilers (#2854)
f6c58c2d2 Make type safe union docs more clear regarding exceptions.
c6c865ba7 Make test threshold a little more robust. 1e-6 was right on the edge and sometimes failed.
be2fa7f93 Add fuse layers for conv+affine+relu and conv+relu (#2842)
efae64281 Work around bug in gcc 11.4.0 causing spurious compile error
9a9b4af6c updated cudnn path (#2843)
ba66152ad Add filter rules for matlab CI
efaa213b6 remove clang 13 CI since the install method isn't working anymore and we have another test for clang
c7b291749 Perform conv with bias for specific algo with cudnn (#2839)
7b6021eee Only trigger CI when relevant files are changed (#2836)
a50bd167c Setup CI for the mex wrapper code (#2834)
f395234aa Update bug.yml (#2835)
3cca5df0f Add sig_traits back since the matlab wrapper depedned on it.
3f9c5c1cf Fix convolution backward filter cache when using cudnn 7 and above (#2828)
7736832ca Optimize pointer to matrix assignment (#2822)
8fc180545 Python: Add grayscale overloads to get_face_chip (#2820)
4e22715eb dlib::optional (#2815)
4d2db66cc Make cblas public (#2817)
18927f93e Fix typos (#2811)
bebba4ca9 Add minimal pyproject.toml (#2804)
759170d3d Fix pixel saturation in interpolate_quadratic (#2806)
eccb9675f [FFmpeg] added save_frame() and improved picking of codecs when not specified (#2800)
b86871241 Remove unneeded using namespace std (#2799)
5b9ab23cc [FFmpeg] added push(image_type) methods to encoder and muxer (#2797)
dab9aa1fa missing include (#2798)
b6418e349 Problem: CUDA error having `code: 716, reason: misaligned address` (#2796)
decdef12f [FFmpeg] decoding and demuxing improvements (#2784)
adf8a3573 put patch level back to 99
39749fba5 tag v19.24.2
e2a0c3dfc std::chrono::system_time on osx isn't in nanoseconds (#2795)
775ac0799 - moved is_detected in type_traits.h (#2794)
191d280d6 [Meta] added callable traits (#2790)
ff331517b [FFT] split FFT API into different layers (#2793)
48fdace27 Use dpoint instead of dlib::vector<double, 2> (#2782)
74a55fb5c FFMPEG : inline docs (#2778)
edc664515 Add insert_image_chip (#2781)
d41571ea5 10X speedup (#2779)
49ac2eb3a Add draw solid polygon (#2775)
53c4be902 FFmpeg : logging (#2774)
5f7026ab5 FFmpeg : improvements to ffmpeg::list() functions (#2773)
b1fe026e0 Add ffmpeg load_frame (#2770)
c21c9c92b Update widgets and draw_string on images to use convert_to_utf32 (#2769)
07f742517 Unify all conversions to UTF-32 (#2737)
89fc01beb FFmpeg : examples (#2764)
20c6364bb Silence warning about unused variable (#2765)
252d76f35 Fix outdated comment in self supervised learning example (#2768)
9148a47c7 RTSP example (#2763)
7155daeee FFmpeg : muxing (#2755)
004b2fe6f Fix imglab changing the current dir too soon (#2761)
4224951c3 Add HSV support (#2758)
689365342 Ubuntu 18.04 Actions runner image is deprecated, use newer image (#2759)
937e07e90 FFmpeg : encoding (#2754)
f586d0a55 FFMPEG : misc + ffmpeg5 support (#2746)
4ffa9b02a Add simple polygon class (#2750)
9a1322997 Exercise more of the ways dlib can be configured in the github actions tests (#2751)
049c59d0f set correct cmake version in check
53d2ceb66 test with cmake 3.8 since that's the min required cmake version for dlib (#2748)
91dcb0405 fix ffmpeg cmake scripts not working with cmake 3.8
4d5db9940 Fix crash when libX11 is missing locale support (#2747)
806b62903 Make everything work in cmake 3.8
b7834324b update sphinx config to work with newer versions of python
15b285131 improve docs
32ab33298 cleanup docs
9d305943b removed extra lines in the MANIFEST.in (#2744)
f89b0388d Add optional Python version (#2743)
93aea6ea9 Update bug.yml
310a37f45 Improve issue template (#2741)
3c0b3620a Added bottom_up_cluster to python api (#2739)
398e2cb5b powerpc/altivec: use __vector instead of vector (#2740)
257b81a17 update docs
ddad576d3 fix wrong union calculation in box_overlap_testing.h (#2735)
eb5de0d53 Add support for loading custom label fonts in imglab (#2733)
dc9475460 Fix for #2729 (#2731)
e006bfe5e Fix typo in input_layer documentation (#2725)
73a1c2f4f Fix unused variable warning in Pybind11 (#2726)
1958da78d Rename ffmpeg examples (#2727)
4519a7a6b Add sum method to running stats (#2728)
50b33753b Strip binaries in release mode (#2721)
2138915df added DLIB_USE_FFMPEG to config.h.in (#2720)
9d3ba472d FFMPEG wrappers: dlib::ffmpeg::decoder and dlib::ffmpeg::demuxer (#2707)
16a8a46a3 Fixed is_nothrow_swappable<> and simplified invoke-related type traits (#2716)
b9ac46816 Added bind_front() and bind_back() (#2705)
1ce7b9cbc Run unit tests with gcc 7 too (#2706)
a12824d42 update example to refer to newer dlib version
2ac3cdbad CMake: Make dependencies private (#2697)
f932f33b1 Fix incorrect xml blocks for leaky_relu and smelu
2efc172c4 removed extra '<' in fc, added '\n' in layers/inputs/losses XML strings (#2698)
02330e0a1 Fix GitHub actions on Windows because of CMake 3.25.0 (#2692)
01c7bfe88 Fix build on ppc64/mips64 (#2689)
4d38ffa08 Suppress incorrect warning from gcc (#2691)
136b2d9e5 Reset last cuda error (#2685)
05d3f6eb1 Disable mac os x python test config since something in the environment is not working properly.
e5b2cedff Improve the data augmentation in the SSL example (#2684)
3d5fb6fc7 Add an informative comment
bdb1089ae Fix computation of the Barlow Twins loss gradient (#2680)
7f06f6e18 Fix empirical cross-correlation computation in the SSL example (#2679)
1bebccfb8 Type erasure tooling (#2666)
dd1ec1fcf Miscellaneous cleanup and refresh (#2661)
20b6c17e7 fix typo in docs
90a8be640 add missing palette_neon_intrinsics.c file (#2664)
19507d4a0 Modernize random_color_transform (#2665)
e50987df4 Fix visitor when tag layer is followed by skip layer (#2662)
65bce59a1 Update Pybind11 to v2.10.0 (#2660)
efe46ce76 Add missing include
4db03ee6f if constexpr() for C++14 (#2654)
902c70193 Do the bias addition in the convolution in a separate call (#2656)
da2f45b2e Add test for libwebp at configure time (#2658)
9b8f5d88f Small YOLO loss improvements (#2653)
bf273a8c2 Add multiclass SVM trainer to svm/auto.h (#2642)
83ec371f1 Use only a fraction of labels for the multiclass SVM in SSL example (#2641)
a5fb8940b - possible for fix for MSVC (#2637)
e2f53a52e stft(), istft(), stftr() and istftr() (#2627)
e2fa5f1e3 Fix Cudnn test project bug (#2634)
69665eb0f Modernize rounding and cast statements (#2633)
89d573645 Only try to find iomp if we have really found the IntelMKL
4867898c5 Make parallel_for_verbose() only print finish statement if it also printed something before then.
4e9854204 Github actions deprecated windows-2016, switching to windows-2019
29288e5d8 Make C++14 and CMake 3.8.0 the new minimum required versions
4de26f144 fix non-determinism in test due to reading uninitialized values.
42a707d75 Make xml parser more robust to bad inputs
f81c4b2d0 YOLO improvements: Smoothed L1 loss, Focal Loss, Weighted Loss (#2628)
6dfba4970 Bug fix for fftr on matrix expression. Clearly nobody is doing FFTs with dlib as this bug has been lingering for a couple years now (#2626)
ad06471a1 Fix typo in the self-supervised learning example (#2623)
270b96c7d oops, add missing deps for some python tests
71d95d76d more tweaks cleanup and fixes for ci builds
c458c4f20 update how python tests are run on ubuntu and mac
135e6ef06 try to fix windows ci testing
a517aaa0b remove broken python tests
febac3adb Try to fix windows python tests
20add2c88 removing travis configs since we haven't used travis in months
1066be80f Update libjpeg to version 9e (#2619)
550a5744d Use chrono in console progress indicator and add finish method (#2617)
3c73978de type_safe_union : visit() and apply_to_contents() performance improvements (#2615)
7c32e1c18 fix spelling (#2614)
ec9a143e0 noexcept dlib::type_safe_union::swap() (#2604)
6c24e65f8 Fix typos (#2606)
7f46535cc Fix potential memory leak when writing WebP throws (#2600)
0f1d8e964 Fix assert having wrong condition
1de792469 Fix saving grayscale WebP images (#2591)
c9a25fda0 update docs
842931975 Record last changeset and set PATCH version to 99
6097093ab Created release v19.24
76a9a1073 update docs
aff6bc440 make sphinx doc building work on newer versions of ubuntu
06b826540 Add SiLU activation layer (#2584)
8af422605 Revert "Add a test on github actions that builds on windows in debug mode"
700908001 Add a test on github actions that builds on windows in debug mode
9db84f4c2 Build ancillary tools on Windows (#2583)
4d92c302c Remove this test on windows for now since it runs out of ram
2bcb1c443 See if using only one thread to build prevents running out of ram on github actions when building windows examples
8ceb91fb4 Do not build WebP features if WebP is not available (#2582)
e264c2664 Build examples and other ancillary tools on windows too
a7ae55b78 Add WebP support to imglab (#2580)
df8b1380a Add GitHub Actions badges (#2571)
ebc4d63d6 Workaround for #2506 (#2570)
a76f205bf Add webp support (#2565)
0aa8b4cbf Treat warnings as errors (#2490)
49314c12d Use CUDA in LayerNorm gradient computation
3a267db57 Fix entry in term_index.html
45f742b82 Update docs
e944ea4fc Update dot visitor
18fe54b64 fix smelu_gradient name
8fa65eb7b Add SmeLU activation
66f9b2b5b Fix documentation
23c74af87 Add directory_exists
9bb1b6f71 Fix focal loss
5cb036564 Make Clang compile cpp11_test.cpp without warnings
e9d49fee8 Print reason for failing to compile cpp11_test
04d1180b0 Make compiler options work with clang-cl
04371157f Increase template-depth for Clang only when needed
4ebf3ef08 Only print Visual Studio messages when compiling with MSVC
3b518f40c Fix deserialize logic in loss_multibinary_log (#2547)
1b7053fe9 Add focal gamma to loss_multibinary_log (#2546)
f1a29f359 Update docs (#2544)
516b744b4 Add missing vistor implementations to visitors.h (#2539)
12f1b3a30 Refactor visitors into their own header (#2533)
736b4931b Fix formatting in YOLO deserialize error message (#2535)
1f9f32ca9 Add visitor to draw nets in the DOT language (#2528)
bf427f56c Problem: `Error while calling cudnnGetConvolutionForwardWorkspaceSize( context(), descriptor(data), (const cudnnFilterDescriptor_t)filter_handle, (const cudnnConvolutionDescriptor_t)conv_handle, descriptor(dest_desc), (cudnnConvolutionFwdAlgo_t)forward_algo, &forward_workspace_size_in_bytes) in file C:\a\2\s\3rdparty\dlib\dlib\cuda\cudnn_dlibapi.cpp:1029. code: 9, reason: CUDNN_STATUS_NOT_SUPPORTED` (#2532)
5f7e19b78 remove cruft
8928f477d Find Threads target if necessary (#2522) (#2526)
0f1b2b37f update cmake docs
efa2baee7 Problem: Error while calling cudnnGetConvolutionBackwardFilterWorkspaceSize( context(), descriptor(data), descriptor(dest_desc), (const cudnnConvolutionDescriptor_t)conv_handle, (const cudnnFilterDescriptor_t)filter_handle, (cudnnConvolutionBwdFilterAlgo_t)backward_filters_algo, &backward_filters_workspace_size_in_bytes) in file C:\a\1\s\3rdparty\dlib\dlib\cuda\cudnn_dlibapi.cpp:1048. code: 9, reason: CUDNN_STATUS_NOT_SUPPORTED (#2527)
1ccd03fec Speed up Barlow Twins loss (#2519)
50b78da53 Fix Barlow Twins loss gradient (#2518)
39852f092 fix race condition in sockets unit test
c7ed14c95 Adjust debugging message from sockets test
f80703e8e Make github actions run the python unit tests (#2517)
c0382b41c Add debugging message to see why the socket tests sometimes fail on windows in github actions
7e942ba98 Update __init__.py.in (#2515)
7be763d65 Make GitHub actions build on gcc-11 and clang too (#2510)
7f8731a24 Add get_random_beta to dlib::rand (#2499)
48f116799 Do not use deprecated `std::random_shuffle` (#2507)
d665bfb89 try to fix build on macOS (#2503)
5291d58e9 See if this makes the MPC test more reliable on MacOS
c89bf6b77 try less parallelism and see if that doesn't break github actions
42c74c38a use parallel builds for github actions
979a22c51 Make Github Actions build more stuff (in particular, the examples) (#2502)
7048f4c1d Avoid build error in visual studio
fda984f7c Add more unit tests for dlib::rand
fc7595d08 Relax MPC test for macOS (#2500)
86b3a809f Make github actions use -q so the test logs are not so spammy
28328af54 Make -q on tests print errors
ffca3b3a6 Add Reorg Layer (#2496)
c91959a73 do not round drectangle in rectangle transform (#2498)
027fb781b GitHub actions (#2289)
3d4c14e63 Add input_rgb_image_pair layer and new net visitors documentation (#2497)
21651f498 Add missing docs for v19.23 (#2495)
2b8625540 Resolve davisking/dlib#2463 (#2492)
54af1f85f Record last changeset and set PATCH version to 99
074ab8bdb Created release v19.23
0cba4bacb updated docs
acf1b9c47 Rename INVOKE to invoke_
56b8cd5cb Make struct some_type; get bolded when pretty printed
8810a62b1 fix spelling error in comment
c1b3feb23 very minor formatting fixup
02b290e3f update release notes
3c654a7a9 cleanup formatting a little
2ed6ae1b6 Eliminate grid sensitivity in YOLO (#2488)
3da3e8118 Fix Layer Normalize (#2489)
aaac87a22 [TYPE_SAFE_UNION] simplified some type traits and added for_each(). (#2475)
42e086960 Fix warning about unused zero_gradients parameter (#2487)
8b4652912 Fix MSVC pragma warnings with other compilers (#2483)
a6f29d41a Avoid different kinds of compiler warnings (#2481)
994df341a add option to not zero out gradients and method to do it (#2477)
a54cea44a added a comment
75b3463fa Avoid distracting compiler warning (#2472)
f0c0b307d Fix crash when truth center is outside of the image (#2471)
a2c476030 Use lambda obj in the ignore case and do some refactoring (#2466)
cf21f5aa9 Add semicolon as separator for reading in a matrix (#2462)
e1ac0b43e normalize samples for SVM classifier (#2460)
44621a612 fix build error
1925b2d10 [INVOKE] added is_invocable_r, invoke_r, made everything constexpr and conditionally noexcept (#2458)
7c8ca2258 minor cleanup
a77b361f8 Make test python builds nice
569de8146 expose momentum_filter to Python (#2457)
5091e9c88 Replace sgd-based fc classifier with svm_multiclass_linear_trainer (#2452)
f77189db0 [INVOKE] C++11 backport of std::invoke, std::invoke_result, std:apply and std::make_from_tuple (#2450)
7f6746e7c Fix nvidia pragma warning (#2454)
709b66677 Avoid redundant computations in the YOLO loss (#2453)
a41b3d7ce We have some excessive and duplicative tests in the travis-ci setup.
2e8bac191 Add dnn self supervised learning example (#2434)
f323d1824 simplification using C++11 types (#2446)
d29a8fc0c Just minor cleanup of docs and renamed some stuff, tweaked formatting.
2b8f9e401 [TYPE_SAFE_UNION] upgrade (#2443)
bf4100069 Fix dnn_trainer trying to decrease the learning rate (#2442)
adca7472d Add support for fused convolutions (#2294)
8a2c74420 Fix trainer with unsupervised loss (#2436)
b9f04fdc4 Fix error in build-and-test.sh script
cd6080ca8 fix spelling error in comment
7b7de0f64 Oops, use correct URI for travis
505e6ed5f update travis uri
e1aa34477 Added mpc option to say you only care about the first time we get to the target
960e8a014 Missing include for `dlib::loss_multiclass_log_per_pixel_` (#2432)
adea4e603 Allow setting custom cuda compute capabilities (#2431)
450697960 add intel mkl search path (#2428)
e0a5725f2 Handle tag layers as inputs to disable_duplicative_biases (#2416)
5bd837d13 Show number of parameters if net is allocated (#2417)
4d7f88bbc Delete .hgtags (#2414)
0a5d5a2c6 updated old message to not say we use mercurial still
7750a6348 produce cleaner python source distribution tarballs
fe0957303 Add progress information to console_progress_indicator (#2411)
74653b4f2 Add function to compute string dimensions in pixels (#2408)
cd915b037 update pngligconf.h
f019b7adc Minor changes to avoid conflicts and warnings in visual studio.
ca3a0fdd5 normalized line endings so visual studio won't complain.
fdf6902ad Another minor thing to avoid warnings from visual studio.
04816ec0f Added missing #include (needed only to avoid gcc warnings)
11101b6f4 update libpng to version 1.6.37
23e506323 update zlib to version 1.2.11
bec25d824 Fix running gradient crashing sometimes (#2401)
16500906b YOLO loss (#2376)
951fdd009 return the projective transform in extract_image_4points (#2395)
b850f0e52 Add LayerNorm documentation (#2393)
e64ea42f6 remove dead code
7d8c6a114 Fix cannot compile iso only code (#579) (#2384)
973de8ac7 Fix disable_duplicative_biases when the input is a skip layer (#2367)
4a51017c2 Make Travis read the CXXFLAGS enviroment variable (#2366)
b99bec580 Fix serialize variant with C++17 (#2365)
9697fa5de [SERIALIZATION] support for std::optional (#2364)
11212a94b [SERIALIZATION] added support for std::variant (#2362)
a54507d81 suppress spurious warning
273d59435 fix comment formatting
cd17f324e fix warnings about possible use of uninitialized values
1de47514b Make input_layer() work with networks that contain repeat layers.
ded68b9af Cleanup gcc version checking code a little.
8e9755ab0 do not attempt to build with gcc 4.8.5 or older (#2357)
9b502d29a Merge branch 'pfeatherstone-allocator_traits'
a051a6542 std_vector.h : making traits compatible with C++20
d6d1a9e87 [TYPE_SAFE_UNION] use std::aligned_union instead of stack_based_memory_block (#2349)
8d4df7c0b remove a unicode character (#2347)
269a3ed1e fix incorrect docs about what gradient is computed
92106718b Make ELU an inplace layer and fix Clipped ReLU implementation (#2346)
7f53d7feb Make clipped-relu inplace and fix docs for elu (#2345)
1b7c7a641 Add Clipped ReLU and ELU activations (#2285)
0ffe9c4c4 Fix input/output mappings with repeat layers (#2337)
a4713b591 Add letterbox image (#2335)
01e6fd51f Record last changeset and set PATCH version to 99
70ea028f1 Created release v19.22
f152a78a5 updated docs
a44ddd745 Add matrix pointwise_pow (#2329)
092fa3037 Add softmax function for matrix type (#2320)
3162f93c5 Revert "Add cmake back as a pip dependency."
0697527ac fix invalid assert
8b9d04390 Add cmake back as a pip dependency.
1b58bdc20 [SERIALIZATION] updated _abstract file (#2306)
479b69e68 Serialization to and from vector<int8_t> and vector<uint8_t> (#2301)
04a3534af fix set_learning_rate_multipliers_range not working (#2304)
42e6ace84 Minor fix in the network size format (#2303)
9f6aefc0d Add missing .get_final_data_gradient() for repeat layer
1516cf31c [MATRIX] bug fix. If you #include <dlib/matrix.h> in a header file you might get a compiler error saying ambiguous call to max(). This commit fixes that. (#2287)
22007f666 work around bug in visual studio
b8807922d FFT: added std::vector overloads for fft, ifft, fft_inplace and ifft_inplace (#2286)
044ff91b7 I'm finding there are other issues with gcc 4.8 now and it's old enough that we can just drop support for gcc 4.8.
95271cfe4 Early termination for find_{min,max}_global (#2281)
600e03655 work around bug in gcc 4.8
869097c80 Make setup.py errors about cmake not being installed more readable
df46c516b Greatly simplify how all the overloads of find_*_global() are setup.
4c0b8b969 Fix compiler warning
0871081a2 fix find_min_global finding wrong answers (#2284)
4a53742f8 Fix build with clang on ppc64le (#2278)
9e5c55674 Fix build errors in gcc 4.8
39eec294b Arbitrary sized FFTs using modified kissFFT as default backend and MKL otherwise (#2253)
71dd9a6cd Remove GUI dependency from fonts (#2273)
187b73f7d Draw string on images (#2270)
74baff132 Fixed CUDA 11.1 compilation. (#2271)
20378b081 make test run faster
0fede57ed Make the mmod loss not spam messages in the unit tests
c5c761b5a Fix build error from ambiguous calls to call_if_valid()
286ca71e1 Release GIL before detecting faces (#2252)
7b5b37502 Update dnn_inception_ex.cpp (#2256)
eb1f08ce6 Removed redundant calls that сause compilation nodiscard warnings. (#2251)
f42a6d23c Some systems manage to install cuda in a way that causes cmake to claim that cuda is available even though it knows it didn't find cublas, which is part of the standard cuda install. So we need to add a check to see if cmake *really for realz* found cuda.
0ff61299b Adding exponential and weibull distributions (#2247)
e58da5135 minor cleanup
d3b021311 Add CIFAR-10 dataset loader (#2245)
d9e58d66c Fixes bug when (de)serializing vector<complex<float>> (#2244)
a7627cbd0 Rename function to disable_duplicative_biases (#2246)
b6bf8aefe Add support for matrix serialization to python API (#2241)
d7644ef2b Expose get_face_chip_details to Python (#2238)
96a75568b fix unused parameter warning in visitor_net_to_xml (#2240)
820fd353d Make dnn_trainer print the minibatch size to ostream (#2236)
375f11722 Add custom ostream to console progress indicator (#2234)
2ef8e3ac1 Update to pybind11 v2.2.4 (closes #1806) (#2229)
93b992d79 slightly improve tests
3f163bd43 Fix pixels being rounded to int values in some cases (#2228)
83921b390 Remove an unused variable and old commented code (#2217)
3c82c2259 Add Layer Normalization (#2213)
507485035 fix backtracking when losses stay at inf (fixes #2206) (#2209)
a1f158379 Do not use sqrt_2 in device code (fixes #2208) (#2210)
3ba004f87 Add GELU activation layer (#2204)
f4f8bff95 fix cmake not finding openblas
c45d166a2 Test cuda losses (#2199)
d78d273a4 Add loss multiclass log per pixel weighted cuda (#2194)
4125a7bb1 DLIB (de)serialization : enhanced STL container support (#2185)
5408b17f7 Linux Distro Detection to fix issue number #2159 #154 (#2169)
0419b8168 Let python users give up to 35 parameters when using the global optimizer.
e7c25c06d Changed directory of license (#2189)
20a147720 update docs
ab346ddfa Extended proxy_(de)serialize objects to work with stringstream, ostringstream, istringstream and vector<char> (#2181)
fa818b9a9 use DLIB_CASSERT to avoid unused variable warning in release compilation (#2182)
d4fe74b5a vectorstream updates: added seekoff and seekpos (#2179)
cdeb2e067 add some docs
12a82f654 Macro for generating default serialisation functions (#2177)
9d60949a3 Add scale_prev layer (#2171)
77e6255fd Add error message for mismatched tensor sizes in dnn_trainer (#2165)
40c3e4881 Simplified more uses of layer visiting and fixed constness bug
5ec60a91c Show how to use the new visitors with lambdas (#2162)
393db2490 switch this to C++11 code
5bcbe617e make type_safe_union movable and also support holding movable types in a natural way.
afe19fcb8 Made the DNN layer visiting routines more convenient.
7dcc7b4eb Added call_if_valid()
e7ec6b777 Add visitor to remove bias from bn_ layer inputs (#closes 2155) (#2156)
ed22f0400 Make dnn_trainer use robust statistic to determine if the loss is exploding and if it should backtrack.
0bb6ce36d dnn_trainer prints number of steps executed when print to ostream
76cc8e3b6 Add probability_values_are_increasing() and probability_values_are_increasing_robust()
c14ba4847 Rename POSIX macro to DLIB_POSIX to avoid name clashes with some libraries.
4b92804dc Use the box with bounding box regression applied to do NMS in the loss.
0e721e5ca Fix bug in bounding box regression loss.
c9809e067 Add missing input/output mappings to mult_prev (#2154)
b401185aa Fix a warning and add some more error handling.
dd06c1169 loss multibinary log (#2141)
d7ca478b7 Problem: With certain batch size / device count combinations, batches were generated with size = 1, causing problems when using batch normalization. (#2152)
bea99ceed switch to a name less likely to conflict with third party code
a9592b07f Minor typo fixes (#2150)
2a870e329 Fix warning for zero variadic macro arguments. (#2151)
f3b4fc548 Added "get_random_complex_gaussian" to dlib::rand (#2149)
f55a1a51a fix python code index page.
59b44849b fix typo, doesn't really matter, but still
02e70ce3c Record last changeset and set PATCH version to 99
9117bd784 Created release v19.21
2e64bdd44 update docs
2c70aad12 Use a cache to avoid calls to the cuDNN algorithm selection routines.
8910445a7 fix some spelling and grammar errors
472107531 More optimization unit tests
a9d554a4a minor cleanup
ff3023f26 Added possibility to load PNG images from a data buffer. (#2137)
c90362d85 updated release notes
7b564927d Switching to what is hopefully a better fix for the following CUDA error
f8cfe6390 Avoid unnecessairly asking cuDNN which algorithms to use, since this is slow in cuDNN 8.0
6c3243f76 Cleanup cuDNN conv algorithm selection code slightly by moving it into its own function.
4d18e0d0c oops, fixing a weird typo
3400e163e tweaked cca test thresholds to avoid false positives
943408d2d Allow forwarding initial function evaluations into find_max_global()
5a80ca9e5 Apply --expt-relaxed-constexpr to all older versions of cuda.
5650ce45a Fix restoration of MSVC warnings in public headers (#2135)
23b9abd07 Switch cuda target architecture from sm_30 to sm_50. I.e. Maxwell instead of Kepler.
a2498dc47 Additional documentation for failed dlib::layer<> use. (#2118)
71b303d88 Improve the error a user gets when trying to build the pyhton extention on windows but they didn't install visual C++.
b9f4da552 Make cuDNN test project failure print a message saying exactly why it failed.
facefa020 Fix random foreset regression not doing quite the right thing.
fe803b566 add support for cudnn 8.0
f8887d8cb fix paths
2a3cb83af Make dlib/__init__.py know the paths to cuda that cmake used. Use those paths instead of relying on CUDA_PATH to be set.
1515adc74 work around a bug in nvcc
7fe355e0e Oops, put import statements after path adjustment
7fcb6c285 Add an __init__.py file so we can add paths to CUDA libs on windows.
53b6ea3bf Record last changeset and set PATCH version to 99
5612caa16 Created release v19.20
610aa6349 updated docs
3d738e1c2 sanitize git logs for export to html
883101477 minor cleanup
a83242014 Corrected interpolate_bilinear for lab_pixel. (#2091)
693aa0a71 fix build errors in cuda 10.2
c7062aa36 Minor optimization: add shortcut to in-place image resize if size_scale is 1 (#2076)
64ba66e1c fix receptive field comment (#2070)
8055b8d19 Update dnn_introduction_ex.cpp (#2066)
253745d29 fix typo in comment
a2e45f00b Reduce code duplication a bit and make equal_error_rate() give correct results when called on data where all detection scores are identical.
0e923cff9 A little bit of cleanup
55e9c890f Add cuda implementation for loss_mean_squared_per_channel_and_pixel (#2053)
b42722a75 Fix DLIB_ISO_CPP_ONLY not working
fbb2db218 fix example cmake script
b44d9465f Fix warning in dnn_trainer initialization list (#2049)
5a715fe24 Remove outdated comment from DCGAN example (#2048)
237746fc1 disable in source builds
e9c56fb21 Fix warnings while running the tests (#2046)
d1d96e380 remove branch from cuda kernel (#2045)
57bb5eb58 use running stats to track losses (#2041)
0057461a6 Promote some of the sub-network methods into the add_loss_layer interface so users don't have to write .subnet() so often.
c79f64f52 make update_parameters() a little more uniform
fd0145345 Fix Use of C++17 deprecated feature: std::iterator #2036
f42f100d0 Add DCGAN example (#2035)
d610e56c2 add leaky_relu activation layer (#2033)
74123841b To avoid a GPU memory leak, allow passing thread pools to dnn_trainer from outside (#2027)
6fc503d24 link against openblasp (#2028)
1380e6b95 add loss multiclass log weighted (#2022)
9185a925c Integer conversions generating compiler warnings (#2024)
08aeada7d Replace result_of by invoke_result for C++17 and above (#2021)
c8a175f56 effect -> affect (#2019)
7b35d7b23 removed inappropriate assert
e7087e595 Prevention of compiler warning (#2015)
c832d3b2f simplify resnet definition by reusing struct template parameter (#2010)
3a53c78ad increment imglab version
9a3366961 A little bit of cleanup and docs. Also added missing mutex lock.
4ff365a53 imglab: chinese ("automatic") clustering, keyboard shortcuts for zooming (#2007)
fc6992ac0 A little bit of cleanup
10d7f119c Add dnn_introduction3_ex (#1991)
c90cb0bc1 Remove unit tests for python 2.7 since that version of pyhton is dead, and the unit test servers don't even support it anymore.
f5c828291 Added note about vcpk availability
b13840a86 Fixed code needing C++14 to use C++11 features instead.
46bcd2059 If nearest-neighbor interpolation is wanted, then don't use an image pyramid. (#1986)
443021882 fix spelling errors (#1985)
870f49a63 Do not link to libnsl (#1987)
20d02b80e run tests for python 3.8 on travis ci
0c415dbb4 Add little test
f71e49f28 remove unused variables
d88b2575a Make copying non-const cuda_data_ptrs to const ones nicer.
bd6994cc6 Add new loss layer for binary loss per pixel (#1976)
6bdd289f7 Added static_pointer_cast() for casting cuda_data_void_ptr to cuda_data_ptr<T>. Also moved some memcpy() functions to namespace scope so that calling them like dlib::cuda::memcpy() can referene them. It was slightly annoting before.
2326a7228 fix code not compiling with some versions of libjpeg as a result of the change I just made.
e1b667181 Fixed const correctness on the in-memory jpeg loading code.
a0af6b7af tweaked docs
60dad52c1 add visitor to count net parameters (#1977)
356bba38f Minor fix: print to console only if the verbose flag is on (#1980)
d766f5e82 Adding Mish activation function (#1938)
a82bf1536 omg, github, cleanup merge stuff
b70c0a6f8 Merge branch 'thebhatman-Mish'
c454bdc18 Added test for Mish
8a91a7c7c Merge branch 'thebhatman-Mish'
edff12d2e Adding Mish activation function
cd5f0b055 fixed failing tests due to recent default change in solver stopping criteria
0c42dcca8 fixed test failure
adce34236 adjusted eps so tests still pass
931fb5265 fixed test not building due to the commit I just made
59d1b9d8c Added a relative epsilon termination option to svm_c_linear_trainer
45731b863 Always check that the data give to cross_validate_trainer() is valid. It's a cheap check, and easy for someone to forget about otherwise.
eae3caf9f Fixed function_evaluation_request::set() invalidating function_evaluation_request::x()
a53354eb7 updated docs
fcc7a75cd Remove pca comment from vector_normalizer documentation (#1965)
54a9a5bbf Fix error for opencv 3.4.9+ over IplImage (#1949) (#1963)
e4998c13b Add sanity check (#1964)
b817bc1ea fixed check range to match the comment
5557577c9 Even the newest CUDA runtime has a buggy cudaStreamSynchronize.
3d5a3d7b9 fixed spelling error
471c3d30e fix formatting
a4bf6e1e6 cleanup cv_image code. This also fixes a build error with the very latest version of OpenCV.
34dc73030 Fix opencv version check to work on all opencv versions
131e45980 Update to work with latest version of OpenCV
2c7e625a1 Record last changeset and set PATCH version to 99
d71497d46 Created release v19.19
591331f94 updated docs
b0e3c3602 deleted old, wrong, and duplicative function docs
e88e166e9 slightly nicer default behavior
f2cd9e3b1 use a time based exeuction limit in example
501d17f69 Made find_max_global() and its overloads measure the execution speed of the user provided objective function. If it's faster than the LIPO upper bounding Monte Carlo model then we skip or limit the Monte Carlo stuff and just fun the objective function more often. Previously, find_max_global() simply assumed the objective functions were really expensive to invoke.
e82e2ceb2 Fix build error in some versions of visual studio
d175c3507 Instance segmentation (#1918)
7f2be82e3 rename top level cmake project to avoid possible cmake issues
bd6dcc0b4 Fix find_max() going into an infinite loop in some cases when a non-differentiable function is given.
1eccfbc99 fix cmake warning
36703f696 Improve check that the system version of libjpeg is usable.
e30f5e2fe Propagate CMake flags to try_compile (fixes #1900) (#1905)
1b83016ab update docs
39327e71b Added note about using cmake's new fetch content feature.
9d3e08647 fix signed/unsigned warning (#1899)
a41e0cdb0 Primarily match to truth rects that haven't been hit already (#1897)
1264521e0 more comments
8e622e9f4 Ignore truth rects that overlap too much (have same index in feature coordinates) (#1896)
cff605add fix warnings
df2bbd8a2 Record last changeset and set PATCH version to 99
1ac175692 Created release v19.18
f4a3d09a9 updated docs and release notes
2e5cf2e8e Add an option to force static runtime (#1847)
e89e28b94 Clear truth_idxs between samples (#1870)
170877da8 add loss_mean_squared_per_channel (#1863)
efd4e2748 dpoint mutates x-coord in y-property (see #1794) (#1866)
a81fd0c65 don't cast away constness (#1865)
e8d240761 Simplified the device_global_buffer() code and API.
d7fb51e60 Fix #1849 by calling device_global_buffer() unconditionally (#1862)
b02be8c73 Const-correct a LAPACK declaration and add aarch64 as a 64-bit architecture (#1859)
7e70a9276 pybind11: cmake: ignore the check between host-python and cross-compiler (#1848)
05cbfc6d6 Fix the CMake BUILDING_PYTHON_IN_MSVC variable not getting picked up where it should.
fced3587f fixing grammar
48b9bf3f9 Fixed cmake warning
8a2963d75 fix mismatch between documentation and implementation (#1835)
2704f3e26 fixed typo in docs
6dec9abff Added missing include
f15e4e58d Fixed grammar in comments
aee97d9d1 Fixed incorrect return type
e2293e5cd Push all include and link options needed for dlib to pkg-config. We do this by getting them from the same list cmake uses.
5d64de24b Fix setting a point's y coordinate changes x instead (Python bindings) (#1795)
22402e99a Revert "Fixed pytest broken dependencies"
e0a24e39f Fix python setup warnings
5e63d0154 Fixed pytest broken dependencies
6de82e241 Added methods for getting keyboard and mouse clicks to image_window's pyhton API.
8001b924e Add input_grayscale_image_pyramid, issue #354 (#1761)
0ecb49b94 Merge branch 'VirgileD-fix/jpeg'
8746c9d37 Stop building parts of libjpeg we don't need.
572302d4d Disable use of non-memory based backing store in libjpeg. This fixes libjpeg not being able to open some types of jpeg file.
a40a5c6e1 Adds a comprehensive error message when jpeg loading fails.
dedb7358b Feature/upgrade libjpeg (#1769)
f0e1d4e47 iEnsures DLIB_FALLTHROUGH macro is only set for GCC>=7 (#1770)
f584e787b Even more warning silencing (#1766)
f3f1a826d more pedantic silencing (#1763)
dbe569a1f fix some pedantic warnings (#1756)
3ce7177a1 Make things work in visual studio.
1fd054baa Suppress compiler warnings
5623810b0 Fixed global_function_search's initialization being wrong if explicitly given an empty list of initial function evaluations.
67d22c48a disable annoying warning
a0d492ae1 Fixed constness for lapack functions (#1737)
879202b79 Fixed build error in newer clang on OpenBSD.
d9af22e53 Merge branch 'master' of https://github.com/davisking/dlib
e1df1968b Include the Intel MKL's iomp dll in the output folder to reduce confusino for windows users.
2dd970a1f Fixed compiler warnings
ccd8b64f1 Semantic-segmentation loss calculation: fix buffer usage on multi-GPU training (#1717)
9433bfd6a Switched to non-deprecated way to say destructor might throw.
64ad6b259 Change CUDA version check for cudaSync from 10000 to 10010 (#1704)
ae406bf4c Make dlib.full_object_detection take list of dlib.point or dlib.points.
b4e7e2e7d fix travis-ci
7e88c5f45 Don't run python tests on travis.
a0dc35edb updated python docs
a4fbe356e Try some os x tests on travis again.
76a65c77b Added option to give --no_ to exclude a unit test from --runall.
eb686ec79 Disable os x python testing for now on travis since something is wrong with the travis config.
d372fba29 try testing python on osx on travis
b8324f5a3 Try to make bash script work for os x.
e3f67173f Try turning on OS X testing on travis.
cc0cbc024 Fixed typo in travis build script.
3401b9693 It's actually fine to call rowm() or colm() with an empty index set.
ec63baa7a Have travis build some tests with DLIB_ASSERT enabled to make sure they work.
f8827f6bb Make test deterministic
89fb5c2e3 cleanup
4e4acf794 Make CMake build a test program to see if libjpeg is really available and not broken.
7611b3476 Set png include paths for test.
0cf422d42 Have CMake build a little test program to see if the system copy of libpng really works.
0268e2748 updated docs
e72d5c71f Exlude unneeded files from python source distributions.
1682c83e8 Removed old mercurial ignore file.
579a41d52 Record last changeset and set PATCH version to 99
810ef7937 Created release v19.17
590a9a11b updated docs
55d575028 updated docs
533e7a374 updated release notes
ffce4af6e Make cusolver finding work for really old versions of cmake with CUDA 10.1
5faf8ccce Automatically test for NEON instructions when building python extensions and enable them if available.
b892df823 Removed --yes option from setup.py since it has long been a noop and its presence just confuses users.
83385e235 Added more dianostic messages from cmake.
663a1b712 Fixed more gcc 8 warnings.
f9b7ad9ae Fixed compiler warnings in gcc 8
5d03b99a0 Changed to avoid compiler warning.
02ed083c4 Work around bug in nvcc in cuda 10.1
c3ee638b3 update ignore list
ba206da01 Switched doc making scripts to pull info from git instead of hg.
70f9a9f11 Made cusolver finding work in newer versions of cmake and cuda
f7f6f6761 Add python api that generates desciptor(s) from the aligned image(s) (#1667)
04a2387cf Make sure CUDA_VERSION is defined (#1663)
23f874d6e Fixed pytest not working with python 2.7. So turning travis tests back on for that version of python.
23c140d13 Removing the python 2.7 test on travis because there is suddenly some kind of incompatibility between python 2.7 and pytest, but only on travis.
16a105691 Removing the install_requires cmake from setup.py because the pip cmake package is hopelessly busted at the moment.
12ca8ad67 merged
9e18e2620 merged
12646178e updated docs
41a87e592 Exposing chinese_whispers directly, closes #1642 (#1644)
84b72278b fixed check for excessive detections in loss_mmod_ (#1625)
ea4519957 Remove Python 3.5 tests on AppVeyor because something is wrong with python 3.5 on appveyor and the other tests are sufficient for python coverage.
ab9ce9984 merged
027e7faf4 Add cmake as a pip install dependency.
5c623e3c9 Apply the #1514 fix even on non-Windows platforms, and change all remaining cudaStreamSynchronize calls (#1596)
9ca72b3e1 Reliably detect AVX2 at runtime on Linux (#1614)
6b5b461dd updated docs
be8dc9269 merged
c763fafde Fixed broken memcpy overload.
f685cb424 Add U-net style skip connections to the semantic-segmentation example (#1600)
fb4c62cc6 remove register keyword from jdhuff for C++17 (#1612)
61a021c93 Add pointwise_divide function (#1586)
99af7b9d5 Fixed code not compiling.
4b2ef5b8f Update CmakeLists: add png for arm (#1580)
e73d3adc0 Add padding argument support for Python compute_face_descriptor methods (#1578)
2cf6b9d55 printing error (#1568)
165d1e356 Add MKL with TBB support to CMake (#1561)
3e9d361f8 Minor changes to avoid bugs in some compilers.
9ff911696 Fix find_blas.cmake bug: find mkl fft on windows. (#1557)
a85a53ce6 See if this unbreaks appveyor
3b9b702c1 Set CMake policy CMP0048 to NEW when available. (#1549)
b54b1c447 merged
f2ad20870 When building python extensions in Visual Studio, never link to any system copies of libpng or libjpeg, instead, use the copies what come with dlib. We are doing this because too many Visual Studio users have busted copies of these libraries on their machines. This will make things just always work.
d7e6f1d72 Apparently fix #1513 by avoiding the cudaStreamSynchronize call (#1514)
d2dcdd66a Fixed random_forest_regression.h warnings (#1544)
ba361d73a Avoid compiler bug
27a529361 Added weighted labels to loss_binary_log layer. (#1538)
08709e8ec Add header needed in some contexts
6747122ca Added --box-images to imglab
9b82f4b0f Record last changeset and set PATCH version to 99
3c3b397f0 Added tag v19.16 for changeset be6f74b9751b
e3cf2349f Created release v19.16
a6ac55ad4 Add error message for users trying to build dlib as a standalone dll in visual studio.
fe10f254f updated release notes
0072afbc7 Python Image and Rectangle Scaling (#1488)
8234080bb merged
ebbc9e896 Fixed compiler warnings
0ab5b70e9 Work around a bug in cmake with regard to cuda 10
65a73ea7a Fix #1479 -- member_function_pointer_kernel_1.h does not compile with MSVC /vms (#1481)
782ade07e ignore more build folders
7b006f376 Added an option to do bounding box regression to the loss_mmod layer.
16c96bc51 updated issue template
6f7db63e5 Fixed typo in docs
603e5ac21 clarified language.
c5ba5d566 A vain attempt to get people to read the manual.
0b597bbe9 updated docs
7d27baf47 updated docs
6ec3aea94 Fixed nvcc getting upset about openmp in some environments.
148dba973 Fixed include order issue. The cuda utils headers weren't including dlib/config.h before the asserts that DLIB_USE_CUDA must be defined. This only matters for users who try to install dlib and use some build system that doesn't define DLIB_USE_CUDA, and so therefore need dlib/config.h to define it for them.
2ed4c6924 Remove redundant line (#1453)
b79a29f1e Fixed is_image_type template always reporting false. This also caused have_same_dimensions() to be uncallable for images, so that's fixed now too.
f95560787 Fixed resize_image() not using the optimized code path in some cases. In particular, it wasn't being used in some parts of the Python API. So this change makes the HOG detector run like 10% faster for instance.
86615775a Test newer versions of pyhton on appveyor.
a54f8e4c3 Made SPD test work more like it used to.
887bc414b Made the non-LAPACK code path in cholesky_decomposition more robust.
62e9cf475 Fixed spelling and tabbing
e3f984a27 merged
4dfd0c82e Allow 64bit seeds for murmur hash.
f32abe975 Slightly changed solver code to avoid a wasted initial iteration.
70a342f44 Fix for "warning C4724: potential mod by 0" in matrix.h (#1424)
98d9703d1 Fixed cblas conflicts in some environments.
4c816be0c Record last changeset and set PATCH version to 99
d353c80ee Added tag v19.15 for changeset 36739b4118e5
929c630b3 Created release v19.15
616a89662 updated docs
d67c3eb31 merged
465e05f5b Work around bug in older version of nvcc.
2dbe70937 merged
d98d1ecc7 merged
71fab625a Added is_image_type and also fixed have_same_dimensions() so that the version specific to images is only called on images. So this fixes a bug where the image version of have_same_dimensions() was being called for tensors and therefore only looking at the nr and nc dimension while ignoring the others. This caused some parts of the DNN tooling to malfunction.
63d963cad merged
a977c03d3 Include the *_tutorial.txt error files in dlib installs.
9848699fb Added static_cast and additional braces to fix compilation warnings. (#1405)
975b6e88e Fixed compile error I just introduced.
7b99544f3 clarified docs
df2142152 Fixed sub_image(), in the pyhton API, not working correctly for non-8bit images.
74042bf65 Cleaned up hysteresis_threshold() code a bit and also removed the hard coded limit on pixel chain recursion depth.
66faa30eb Record last changeset and set PATCH version to 99
2caee4699 Added tag v19.14 for changeset 20171d55ebbc
62ad0416a Created release v19.14
f0e6a0aff updated docs
1b70c9f92 clarified spec
a25f47a3e updated docs
20130eb1f Clarified docs
bca6939fe Clarified docs
3f19503a3 Split image2.cpp into two files to help speed up the build.
4aed5b2b7 Split image.cpp into two files so builds are more parallelizeable.
ea2fbcab5 Disable LTO since it basically just makes the build process slow.
e248cc8fb Added spatially_filter_image() and spatially_filter_image_separable() do the python API.
93b83677a clarified docs
0f169ed71 Just a bit of code cleanup
ee5bb2a34 Made rectangle_transform measure area the same way the rectangle's .area() does so that things are a little more consistent.
5fa275f89 Made the routines that rotate image datasets use the rectangle_transform instead of the bad old way that only really worked for square boxes.
112d32afc Fixed spelling errors in comment.
e6b529625 merged
e552afe9e merged
da200b22c merged
18711b55a detect file extension of mex function
1a1ec6881 Fixed errors on OS X for some kinds of builds.
3b794540b Make pybind11 not break builds that include cuda files.
3de8fa368 Set visibility to hidden for cuda as well as c++.
a60be3415 Fixed busted assert.
8fdb3af9b Fixed docstrings not showing the right types.
d96d3a50f cleanup
ec342e39b Allow batched face recognition for greater performances (#1335)
eded3773f Never use anaconda's version of libpng because it's broken.
6b581d91f Added detection_window_width, detection_window_height, and num_detectors fields to the simple object detection objects in the Python API.
8b5ee898a Made overlay_line use dpoints instead of point and also made it render the line from the middle of the pixel.
92a89bceb Allow point and dpoint use for hough_transform stuff.
746af7f86 Made the hough_transform's functions for getting line coordinates use dpoint instead of point.
cd8411a8a Made clip_line_to_rectangle() and nearest_point() work for point and dpoint.
9330c9736 fixed broken test
9a9eb5e53 Clarified docs
49e6002df Fixed hough_transform::find_pixels_voting_for_lines(). It would sometimes include a pixel multiple times in the output lists.
d0fc6023c Updated count_points_on_side_of_line() python binding.
0c652dd46 Made count_points_on_side_of_line() take a lower threshold in addition to the upper threshold. This changes the function's API.
5f9c881f7 Made remove_duplicates() take any type rather than just being limited to rectangle.
64fb23124 Fixed grammar in docs
cf8f5581b Fixed grammar in comment.
d2efb6893 Made hysteresis_threshold() return a uint8 image so it's consistent with threshold_image().
3aad318e9 Made add_overlay() work with more input types in python.
dcec88672 Fixed integer overflow.
527e26df0 Added max runtime option to object detector trainer API in python.
7d3fac550 Added a --split-train-test option to imglab.
5cea44a28 Added a max runtime option.
918bbd5d8 Added a max runtime option to the oca solver.
d3006ab3b Made it so you can call the std::vector<detectors> version of the object_detector constructor from python. So now you can pack multiple detectors into one object via the python API.
d50d8e026 Made the run_multiple() routine work identically for fhog and simple object detectors.
19005f68c Added translate_rect() to python API
165310c51 Fixed print formatting.
c255e73ba Made it so you can display arbitrary full_object_detection objects on the image_window in Python.
8993a0d9d Changed the overlay_circle so you can give a double radius rather than just an integer radius. Also made the circles centered on the center of the pixel at their center when you zoom in.
4b907532d Allow python to set the padding mode of the shape_predictor_trainer.
8798955d5 Made it so you can fully build image_dataset_metadata::dataset objects from python.
30ba39c9a Fixed cblas_dtrmm and cblas_strsm not using CBLAS_INT_TYPE.
85a0dea37 Added zero_border_pixels() to the python API
1429b8678 Fix cuda not building on osx sometimes.
1e8c2cc1b Removed spurious ;
4f7c01f46 Added more overloads of memcpy() for cuda_data_ptr.
b23c71864 Fixed dlib.threshold_image() ignoring the optional thresh argument.
1f0cb544c Added max_point() and max_point_interpolated() to the Python API.
f8419124c Cleaned up the object_detector Python API a little and also pushed the nuclear norm regularization stuff to Python. This also includes adding num_separable_filters() and threshold_filter_singular_values() to the Python API.
9a845c511 A bit of cleanup
5ed1c2f28 Made it so you can construct dlib.image_windows from detectors.
653d04837 Test for SSE4 instructions on the host before enabling them in the python extension module.
dcebe3391 Fixed typo in comment string.
1d6714951 Added CMakeLists.txt for the stuff in tools/archive.
3126372c1 Added the translation jittering option to the python API.
a9c940b16 Added an option to do translational jittering of the bounding boxes in the shape_predictor_trainer.
89bfb786d Fixed be_verbose not doing the right thing.
0fe68bb5a Added auto_train_rbf_classifier() and reduced() to the Python API.
2a27b690b Added auto_train_rbf_classifier()
c14dca071 Added an optional thread_pool argument to find_max_global() and find_min_global() to allow for parallel execution of the objective function calls.
66c4fa81c Made decision functions more viewable from python.
106c905b7 Added a comment
e3c78bcd6 Clarified message
048cb4ad1 merged
64822a063 Ignore bmp reserved area (#1343)
908f5540a Make arrays of points and rectangles constructable with a size in the Python API.
44c06150a Fixed numpy_image not compiling when C++14 is enabled.
85fc0e0b9 Added a have_same_dimensions() for image types.
490754b30 Fix bad_alloc exception when loading bmp images with negative height (#1341)
4fec20c47 Added option to link against mkl_sequential (#1302)
5dc918581 updated docs
f24b0816a Made .size of hough_transform a property rather than a function.
212c19c0c cleanup
18d438588 Fixed spelling error in comment
0d6f1f75d Gave numpy_image a (rows,cols) constructor.
d6b1f69b5 Cleaned up numpy_image code a little.
b6267d7e6 Made the image view objects implement the generic image interface. Not sure why I didn't do this a long time ago.
771036cee Cleanup hough_transform pyhton binding
996a4bbba Fixed cmake erroring out in some situations.
e27fcb2a2 Print version information before doing anything else in cmake.
bf815eea3 Fixed spelling error
b9146ad67 Fixed spelling error
06bbfcb44 Added polygon_area() to python API
263e64ddf Added polygon_area()
a40614e58 Added sub_image() to python API. Also added an overload of get_rect() for hough_transform objects and made some argument names more consistent.
342657090 More overloads for rectangle.contains()
4f2f565cb Added more operators for dpoint and point in the Pyhton API.
81d57e72c Cleaned up min_barrier_distance() code a bit and also made it more robust.
0249bcb9d Added get_histogram() to the python API. Also added more overloads of label_connected_blobs().
5b32723fd Avoid implicitly copying a matrix to numpy_image.
fff972347 Fixed label_connected_blobs_watershed() not giving contiguous labels as outputs. Also cleaned up the code a bit.
1a3f8d5bf Added center() and more overloads of centered_rect() to the Python API.
8516442ae Added extract_image_chip() and extract_image_chips() to the Python API.
c210395c0 Made chip_details do something more benign when the user gives an empty rectangle.
2ed128e2b Made dlib.range a little more like the python range.
d2fa9e85e Cleaned up code and fixed point and dpoint addition actually doing subtraction.
bef30d183 merged
8402c9ee9 make pypi not complain about long description.
8cfeb4692 Record last changeset and set PATCH version to 99
c9e2f040f Created release v19.13
01e14af26 Added tag v19.13 for changeset c7a90ee40635
4f159e7e5 updated docs
35d6a5b76 Added images for docs
57e4fd207 Added a .add_overlay_circle() to dlib.image_window.
91e64dfb1 Added min_barrier_distance() to the Python API.
e1458ec8e Added min_barrier_distance() and also a new overload of get_histogram().
233ae95c1 Cleaned up code and pushed is_rgb_image and is_grayscale_image somewhere more accessible.
37d13ae88 Fixed some python functions not taking as wide a range of images as they did in previous dlib versions.
027cf6434 Added an assign_image() that takes py::array as source image.
bd2e0bc30 Made skeleton(), in the python API, return the given image so you can chain calls easily.
61687a8f0 merged
c1f5f32b2 Renamed convert_rgb_to_grayscale() to as_grayscale() and made it take any numpy array as input.
89ffae612 Added load_grayscale_image() to the Python API
840bb5bc6 Updated to work with new extract_image_4points() interface.
91ae3de1c Made extract_image_4points() use std::array instead of std::vector and also made it more robust to funky inputs.
e3a9ddb2e Added python_list_to_array()
013dd0177 Added is_convex_quadrilateral(), find_convex_quadrilateral(), and no_convex_quadrilateral.
f82aa1bee Add a default cpack "package" target. (#1330)
d3bd9d61e Added extract_image_4points() to the Python API.
7eb1d2d4d Added angle_between_lines() to the Python API.
8785ac1d9 Clarified spec
844b35523 Added extract_image_4points()
5edab25c5 Added angle_between_lines()
ec150f8d8 Fixed warnings and sorta bugs.
f2cb001c8 Improved error messages.
39a132983 Added a version of transform_image() that works with projective transforms to the Python API.
a67bdd8d7 Added point_transform_projective, find_projective_transform(), and inv() to the Python API.
a18e72e51 A bit of cmake cleanup
ec83abf61 Removed notes about using --yes USE_AVX_INSTRUCTIONS when building python interface since AVX availability is now detected automatically by cmake.
104ffddb4 Make python extension automatically enable AVX instructions if the host machine supports them.
f21ac0806 Added hough_transform, remove_incoherent_edge_pixels(), normalize_image_gradients(), line, signed_distance_to_line(), distance_to_line(), reverse(), intersect(), count_points_on_side_of_line(), count_points_between_lines(), dot(), and normalize() to Python API.
ecde075d4 Added an image_window::add_overlay() overload for line object.
08e06b120 updated spec
6472306f2 Fixed docs
67bbd2559 Clarified spec and fixed typo
a190a1c24 Added equalize_histogram() and resize_image() to the Python API.
1df5227e3 Added hysteresis_threshold() and sobel_edge_detector() bindings in the Python API.
5df921a22 Made the python API for image_window accept any image type.
a8dc28ad1 Fixed code not compiling in gcc 4.8
a3ca78de9 Added convert_rgb_to_grayscale(), pyramid_down, find_bright_keypoints(), find_bright_lines(), find_dark_lines(), find_dark_keypoints(), suppress_non_maximum_edges(), and find_peaks() to the Python API.
40da8bf72 Added length() for point and dpoint in the Python API.
52e7ab68b Made image_window use numpy_image explicitly so it benefits from numpy_image's automatic binding.
cafd6a0ad Added centered_rect() and centered_rects() to Python API. Also added the *_corner() routines to rectangle and drectangle and made these objects constructable from instances of each other.
c18202842 Made point and dpoint constructable from each other.
f917f5e5f Made the numpy_image capable of binding to tuples that contain a valid image as their first element.
1619725f2 Improved input validation for numpy_image.
e83f35039 Fixed numpy_image and pybind11 crashing python sometimes when certain types of conversions are attempted.
86a9abe09 fixed typo in docs
70c6f7b84 Minor cleanup
c0097a782 Added dpoint to the python API.
da943c691 Added label_connected_blobs_watershed(), convert_image(), and convert_image_scaled() to the python API.
eca48c346 clarified spec
baaec4a52 Fixed label_connected_blobs_watershed() only working in C++14 and if all parameters were given.
e15fcbd39 Added the following things to the python API: gaussian_blur(), label_connected_blobs(), randomly_color_image(), jet(), skeleton(), find_line_endpoints(), get_rect(), shrink_rect(), grow_rect(), and image_gradients.
28283a162 Fixed grammar
827dd0f81 updated pypi upload instructions.
dddfb6504 Record last changeset and set PATCH version to 99
bb3dc1752 Added tag v19.12 for changeset 39949a24fa62
a1d1be69f Created release v19.12
b19097ed5 updated docs
f5a7faa54 updated faq
7ae59ff82 updated docs
26701f734 Fixed typo in docs
1f727f930 Fixed broken unit test.
7e7d8e907 Added python interface to threshold_image() and partition_pixels().
27dbbe2de Renamed dlib.save_rgb_image() to dlib.save_image()
68112e350 Improved the error messages related to numpy_image usage and also improved the way overload resolution works in pybind11 for these objects.
06ce20785 Fixed numpy_image not working correctly for grayscale images.
2b55b9964 Record last changeset and set PATCH version to 99
0feb4de1d Added tag v19.11 for changeset 4ae153db78c4
62ca29f3b Created release v19.11
7f61b773e updated docs
e90fc1076 clarified spec
ba4486363 Fixed grammar
d014a0be8 updated change for last release.
0e0675715 Improved python docs
43b9a0966 Added encode_8_pixel_neighbors() and find_line_endpoints().
b2f6071a9 Changed the matrix_data_layout types to use size_t where appropriate so that large matrices can be handled on windows.
150766bb5 Made memory managers use size_t to measure memory allocation sizes rather than unsigned long.
ab946d7fc Added find_bright_keypoints() and find_dark_keypoints(). Also added a convenience overload for remove_incoherent_edge_pixels().
f2845f15a Added find_peaks()
5a2e0d6a4 Added centered_rects()
b31780590 Added overload of hysteresis_threshold() that uses partition_pixels() to select thresholds.
e340f1cb0 Added an overload of partition_pixels() that provides multiple output thresholds.
69a53cc9f updated docs
92e115810 Added label_connected_blobs_watershed()
2673694cd Fixed cell arrays of complex types not binding correctly.
a6abf8c6f Added image_gradients documentation images.
ee966b641 updated docs
8b7cb1f50 clarified spec
cf4f109ec Added image_gradients as well as find_bright_lines() and find_dark_lines().
a79b72c58 Added paper
6bdfe3cd1 Make sure -DDLIB__CMAKE_GENERATED_A_CONFIG_H_FILE is propagated to nvcc so that the odr violation checks don't erroneously trigger.
5fe116d2d Made partition_pixels() work basically all pixel types.
69ee698da Fixed bug in hysteresis_threshold()
d5a27eba6 updated docs
802d64110 Added partition_pixels() and made threshold_image() use it to find the default threshold if none is given. Also depreciated auto_threshold_image() since using partition_pixels() to pick the threshold is generally better.
940546f87 Make python docs include __call__ and __init__
851d70a5b A bit of cleanup
a37284fd4 Flush to stdout so that print messages appear where you would expect relative to output from the subprocesses.
c5f5692a3 more appveyor stuff
627ce87d9 Fix appveyor tests
56ce7bc5d Build clean on appveyor.
5a3684263 Print python version in build log.
f4dc4855c Cleanup
d847c1301 Added neighbors_24
8a48eee40 Added more version status messages to cmake output.
1a297363c Fixed spelling error in message.
7eac06947 Cleanup cmake output a little.
4b44c2a38 Fix warnings on OS X.
0ff10dcc7 fixed include
cd888bcc3 merged
460798622 parallelize image rotations (#1281)
ec3c36fc2 Set execute bit.
46052db07 Turn LTO back on.
6980697a4 Just moved code into dlib namespace.
43e5f42ef Added numpy_image templated class that makes a numpy array conform to dlib's generic image interface. This makes dealing with numpy images in pybind11 modules much nicer.
f6651c2c7 Fixed jitter_image() not following the generic image interface.
58888fcbd Merged
b95cb7d4a Added casts to cblas calls to avoid conflicts with some definitions of the cblas API.
69b3894f4 Supporting bottom-up DIBs (#1277)
4d793ddd8 Added some utility routines to the hough_transform object.
26c2cc4e4 updated docs
448473650 Added line unit tests.
f72537616 Added the line class an some related utility functions.
c56eaf0a0 Added normalize_image_gradients() and remove_incoherent_edge_pixels().
3c052e8e9 Fixed code not compiling in debug mode.
04eb33ce0 Added find_pixels_voting_for_lines() to the hough_transform.
4f29050a6 Test against python 3.6 on windows.
e41fad6b6 Added a convenience overload to hough_transform.
b5988db6a Added .begin() and .end() to array2d.
30101dff4 Updated about to talk about new part labeling method.
f2167cb6b Made it so shift clicking on a selected rectangle in the image_display will add a part with an automatically selected integer label if and only if you haven't specified any string part labels and all the existing parts are already integer labeled. This makes it way easier to quickly add parts.
1184e9381 Renamed BOOST_JOIN to DLIB_BOOST_JOIN to prevent name clashes when working with boost (#1265)
52322e5ae Made convert_to_numpy() take an rvalue reference so that it's really explicit that it's going to move the matrix into the result.
905074f2c updated patch version to 99
e8faced82 Add basic image io and remove python C-API refs from numpy_returns.cpp (#1258)
d7dfd8ad2 Made find_min_single_variable() more stable. Fixes https://github.com/davisking/dlib/issues/1224.
854801d4b Improved assert error message.
998853562 Fixed compile time error in the matlab bindings.
4104c2cd1 Updated windows test bat file.
59483b28c merged
9b040cd7c Added steal_memory() to the matrix.
74a3950c1 fixed indenting.
509d62351 Fixed refcount issue in Python dlib.jitter_image and dlib.get_face_chips (#1246)
24ac9bc43 A bit of cmake cleanup
4a97b216e updated docs
592361ed8 Moved the CUDA stuff into its own folder.
49a5d39d2 Made loss_multiclass_log_per_pixel use CUDA.
6c962dd92 Fixed spelling error in comment.
a92477b99 Improved cuda_data_ptr.
bb8e0bc8b More cleanup
0ce6ed5be Cleanup of cuda code.
8073f4b16 Make visual studio build with all cores when building dlib.
aba6e3639 A bit of cleanup
51101fa95 prevent C++ compilation flags leaking to CUDA (#1216)
fe1c28737 Remove unnecessary semicolon (#1223)
0db8713a6 Fixed reference count issue (#1222)
d36c66197 Still messing with travis
28418216d more travis debug
edfd510d7 more travis debug
ec72f92f9 more travis debug
3787be84e more travis debug
6108f60aa more travis debug
526fa5023 more travis debug
963bc0fea merged
9f53eac2b Improved the stopping condition of solve_qp_box_constrained() and solve_qp_box_constrained_blockdiag().
973a3da36 more travis debug
5597679bb more travis debug
f5665cd15 more travis debug
613ac8906 more travis debug
81248d3f9 more travis debug
9bf9340d4 debug travis
fe7eb3a06 debug travis
11020cbdc Debug travis
f302ab6fd Try to fix travis
8587ccd0b Added logging
d0b94a76c Fixed cmake downloading sc…
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
When trying to refer to mitie.total_word_feature_extractor by importing mitie.py from latest python 2.7 binary, getting an error
AttributeError: 'module' object has no attribute 'total_word_feature_extract
mitie.py in binary version is missing the function total_word_feature_extract. Looks like the binary needs to be refreshed to reflect changes in master. Is it possible to upload the latest python binary?
The text was updated successfully, but these errors were encountered: