75bcd397af
* KinFu demo: idle mode added * undistort added * KinFu demo: depth recorder added * TSDFVolume gets voxelSize, voxelSizeInv, truncDist members; decorative fixes * TSDFVolumeGPU::integrate(): host code compiles * TSDFVolume: truncDist fixed * TSDFVolume::integrate(): initial OCL version is done * TSDFVolume::integrate(): OCL: minor fixes * kinfu: small fixes * TSDFVolume::raycast(): initial GPU version is done * USE_INTRINSICS directive for centralized enable/disable opt. code * TSDF Volume supports 3 different sizes/resolutions on each dimension * TSDFVolume: serviceMembers moved to parent class * TSDFVolumeGPU: mem order changed, gave perf boost 4x * Frame: fixed UMat as InputArray; TSDF: comments, TODOs, minor fixes * Frame::getPointsNormals(); minors * FrameGPU: initial version (not working) * minor * FrameGPU: several fixes * added OCL "fast" options * ICP OCL: initial slow version is done (host-side reduce) * ICP OCL: reduce is done, buggy * KinFu Frame: more args fixes * ICP OCL: small fixes to kernel * ICP OCL reduce works * OCL code refactored * ICP OCL: less allocations, better speed * ICP OCL: pose matrix made arg * Render OCL: small fix * Demo: using UMats everywhere * TSDF integrate OCL: put const arg into kernel arg * Platform parameter partially removed, implementation choice is done through OCL availability check * Frame and FrameGenerator removed (other code is in following commits) * CPU render: 4b instead of 3b * ICP: no Frame class use, one class for both CPU and GPU code * demo: fix for UMat chain * TSDF: no Frame or FrameGenerator use * KinFu: no Frame or FrameGenerator, parametrized for Mat or UMat based on OCL availability * KinFu::setParams() removed since it has no effect anyway * TSDF::raycast OCL: fixed normals rendering * ScopeTime -> CV_TRACE * 3-dims resolution and size passed to API * fixed unexpected fails of ICP OCL * voxels made cubic again * args fixed a little * fixed volSize calculation * Tests: inequal, OCL, unified scene test function * kinfu_frame: input types fixed * fixed for Intel HD Graphics * KinFu demo: setUseOptimized instead of setUseOpenCL * tsdf: data types fixed * TSDF OCL: fetch normals implemented * roundDownPow2 -> utils.hpp * TSDF OCL: fetch points+normals implemented * TSDF OCL: NoSize, other fixes for kernel args * Frame OCL: HAVE_OPENCL, NoSize, other kernel args fixed * ICP OCL: HAVE_OPENCL, NoSize and other kernel fixes * KinFu demo fixes: volume size and too long delay * whitespace fix * TSDF: allowed sizes not divisable by 32 * TSDF: fixed type traits; added optimization TODOs * KinFu made non-free * minor fixes: cast and whitespace * fixed FastICP test * warnings fixed: implicit type conversions * OCL kernels: local args made through KernelArg::Local(lsz) call * MSVC warnings fixed * a workaround for broken OCV's bilateral * KinFu tests made a bit faster * build fixed until 3.4 isn't merged to master * warnings fixed, test time shortened
Repository for OpenCV's extra modules
This repository is intended for development of so-called "extra" modules, contributed functionality. New modules quite often do not have stable API, and they are not well-tested. Thus, they shouldn't be released as a part of official OpenCV distribution, since the library maintains binary compatibility, and tries to provide decent performance and stability.
So, all the new modules should be developed separately, and published in the
opencv_contrib repository at first. Later, when the module matures and gains
popularity, it is moved to the central OpenCV repository, and the development team
provides production quality support for this module.
How to build OpenCV with extra modules
You can build OpenCV, so it will include the modules from this repository. Contrib modules are under constant development and it is recommended to use them alongside the master branch or latest releases of OpenCV.
Here is the CMake command for you:
$ cd <opencv_build_directory>
$ cmake -DOPENCV_EXTRA_MODULES_PATH=<opencv_contrib>/modules <opencv_source_directory>
$ make -j5
As the result, OpenCV will be built in the <opencv_build_directory> with all
modules from opencv_contrib repository. If you don't want all of the modules,
use CMake's BUILD_opencv_* options. Like in this example:
$ cmake -DOPENCV_EXTRA_MODULES_PATH=<opencv_contrib>/modules -DBUILD_opencv_legacy=OFF <opencv_source_directory>
If you also want to build the samples from the "samples" folder of each module, also include the "-DBUILD_EXAMPLES=ON" option.
If you prefer using the gui version of cmake (cmake-gui), then, you can add opencv_contrib modules within opencv core by doing the following:
-
start cmake-gui
-
select the opencv source code folder and the folder where binaries will be built (the 2 upper forms of the interface)
-
press the
configurebutton. you will see all the opencv build parameters in the central interface -
browse the parameters and look for the form called
OPENCV_EXTRA_MODULES_PATH(use the search form to focus rapidly on it) -
complete this
OPENCV_EXTRA_MODULES_PATHby the proper pathname to the<opencv_contrib>/modulesvalue using its browse button. -
press the
configurebutton followed by thegeneratebutton (the first time, you will be asked which makefile style to use) -
build the
opencvcore with the method you chose (make and make install if you chose Unix makfile at step 6) -
to run, linker flags to contrib modules will need to be added to use them in your code/IDE. For example to use the aruco module, "-lopencv_aruco" flag will be added.
Update the repository documentation
In order to keep a clean overview containing all contributed modules the following files need to be created/adapted.
-
Update the README.md file under the modules folder. Here you add your model with a single line description.
-
Add a README.md inside your own module folder. This README explains which functionality (seperate functions) is available, links to the corresponding samples and explains in somewhat more detail what the module is expected to do. If any extra requirements are needed to build the module without problems, add them here also.