v0.21.0

• build
  • added go clean --cache to clean target, see issue 458
• core
  • Add KMeans function
  • added MeanWithMask function for Mats (#487)
  • Fix possible resource leak
• cuda
  • added cudaoptflow
  • added NewGpuMatFromMat which creates a GpuMat from a Mat
  • Support for CUDA Image Warping (#494)
• dnn
  • add BlobFromImages (#467)
  • add ImagesFromBlob (#468)
• docs
  • update ROADMAP with all recent contributions. Thank you!
• examples
  • face detection from image url by using IMDecode (#499)
  • better format
• imgproc
  • Add calcBackProject
  • Add CompareHist
  • Add DistanceTransform and Watershed
  • Add GrabCut
  • Add Integral
  • Add MorphologyExWithParams
• opencv
  • update to version 4.1.2
• openvino
  • updates needed for 2019 R3
• videoio
  • Added ToCodec to convert FOURCC string to numeric representation (#485)

v0.20.0

• build
  • Use Go 1.12.x for build
  • Update to OpenCV 4.1.0
• cuda
  • Initial cuda implementation
• docs
  • Fix the command to install xquartz via brew/cask
• features2d
  • Add support for SimpleBlobDetectorParams (#434)
  • Added FastFeatureDetectorWithParams
• imgproc
  • Added function call to cv::morphologyDefaultBorderValue
• test
  • Increase test coverage for FP16BlobFromImage()
• video
  • Added calcOpticalFlowPyrLKWithParams
  • Addition of MOG2/KNN constructor with options

v0.19.0

• build
  • Adds Dockerfile. Updates Makefile and README.
  • make maintainer tag same as dockerhub organization name
  • make sure to run tests for non-free contrib algorithms
  • update Appveyor build to use Go 1.12
• calib3d
  • add func InitUndistortRectifyMap (#405)
• cmd
  • correct formatting of code in example
• core
  • Added Bitwise Operations With Masks
  • update to OpenCV4.0.1
• dnn
  • add new backend and target types for NVIDIA and FPGA
  • Added blobFromImages in ROADMAP.md (#403)
  • Implement dnn methods for loading in-memory models.
• docker
  • update Dockerfile to use OpenCV 4.0.1
• docs
  • update ROADMAP from recent contributions
• examples
  • Fixing filename in caffe-classifier example
• imgproc
  • Add 'MinEnclosingCircle' function
  • added BoxPoints function and BorderIsolated const
  • Added Connected Components
  • Added the HoughLinesPointSet function.
  • Implement CLAHE to imgproc
• openvino
  • remove lib no longer included during non-FPGA installations
• test
  • Add len(kp) == 232 to TestMSER, seems this is necessary for MacOS for some reason.

v0.18.0

• build
  • add OPENCV_GENERATE_PKGCONFIG flag to generate pkg-config file
  • Add required curl package to the RPM and DEBS
  • correct name for zip directory used for code download
  • Removing linking against face contrib module
  • update CI to use 4.0.0 release
  • update Makefile and Windows build command file to OpenCV 4.0.0
  • use opencv4 file for pkg-config
• core
  • add ScaleAdd() method to Mat
• docs
  • replace OpenCV 3.4.3 references with OpenCV 4
  • update macOS installation info to refer to new OpenCV 4.0 brew
  • Updated function documentation with information about errors.
• examples
  • Improve accuracy in hand gesture sample
• features2d
  • update drawKeypoints() to use new stricter enum
• openvino
  • changes to accommodate release 2018R4
• profile
  • add build tag matprofile to allow for conditional inclusion of custom profile
  • Add Mat profile wrapper in other areas of the library.
  • Add MatProfile.
  • Add MatProfileTest.
  • move MatProfile tests into separate test file so they only run when custom profiler active
• test
  • Close images in tests.
  • More Closes in tests.
  • test that we are using 4.0.x version now
• videoio
  • Return the right type and error when opening VideoCapture fails

v0.17.0

• build
  • Update Makefile
  • update version of OpenCV used to 3.4.3
  • use link to OpenCV 3.4.3 for Windows builds
• core
  • add mulSpectrums wrapper
  • add PolarToCart() method to Mat
  • add Reduce() method to Mat
  • add Repeat() method to Mat
  • add Solve() method to Mat
  • add SolveCubic() method to Mat
  • add SolvePoly() method to Mat
  • add Sort() method to Mat
  • add SortIdx() method to Mat
  • add Trace() method to Mat
  • Added new MatType
  • Added Phase function
• dnn
  • update test to match OpenCV 3.4.3 behavior
• docs
  • Add example of how to run individual test
  • adding instructions for installing pkgconfig for macOS
  • fixed GOPATH bug.
  • update ROADMAP from recent contributions
• examples
  • add condition to handle no circle found in circle detection example
• imgcodecs
  • Added IMEncodeWithParams function
• imgproc
  • Added Filter2D function
  • Added fitLine function
  • Added logPolar function
  • Added Remap function
  • Added SepFilter2D function
  • Added Sobel function
  • Added SpatialGradient function
• xfeatures2d
  • do not run SIFT test unless OpenCV was built using OPENCV_ENABLE_NONFREE
  • do not run SURF test unless OpenCV was built using OPENCV_ENABLE_NONFREE

v0.16.0

• build
  • add make task for Raspbian install with ARM hardware optimizations
  • use all available cores to compile OpenCV on Windows as discussed in issue #275
  • download performance improvements for OpenCV installs on Windows
  • correct various errors and issues with OpenCV installs on Fedora and CentOS
• core
  • correct spelling error in constant to fix issue #269
  • implemented & added test for Mat.SetTo
  • improve Multiply() GoDoc and test showing Scalar() multiplication
  • mutator functions for Mat add, subtract, multiply, and divide for uint8 and float32 values.
• dnn
  • add FP16BlobFromImage() function to convert an image Mat to a half-float aka FP16 slice of bytes
• docs
  • fix a varible error in example code in README

v0.15.0

• build
  • add max to make -j
  • improve path for Windows to use currently configured GOPATH
• core
  • Add Mat.DataPtr methods for direct access to OpenCV data
  • Avoid extra copy in Mat.ToBytes + code review feedback
• dnn
  • add test coverage for ParseNetBackend and ParseNetTarget
  • complete test coverage
• docs
  • minor cleanup of language for install
  • use chdir instead of cd in Windows instructions
• examples
  • add 'hello, video' example to repo
  • add HoughLinesP example
  • correct message on device close to match actual event
  • small change in display message for when file is input source
  • use DrawContours in motion detect example
• imgproc
  • Add MinAreaRect() function
• test
  • filling test coverage gaps
• videoio
  • add test coverage for OpenVideoCapture

v0.14.0

• build
  • Add -lopencv_calib3d341 to the linker
  • auto-confirm on package installs from make deps command
  • display PowerShell download status for OpenCV files
  • obtain caffe test config file from new location in Travis build
  • remove VS only dependencies from OpenCV build, copy caffe test config file from new location
  • return back to GoCV directory after OpenCV install
  • update for release of OpenCV v3.4.2
  • use PowerShell for scripted OpenCV install for Windows
  • win32 version number has not changed yet
• calib3d
  • Add Calibrate for Fisheye model(WIP)
• core
  • add GetTickCount function
  • add GetTickFrequency function
  • add Size() and FromPtr() methods to Mat
  • add Total method to Mat
  • Added RotateFlag type
  • correct CopyTo to use pointer to Mat as destination
  • functions converting Image to Mat
  • rename implementation to avoid conflicts with Windows
  • stricter use of reflect.SliceHeader
• dnn
  • add backend/device options to caffe and tensorflow DNN examples
  • add Close to Layer
  • add first version of dnn-pose-detection example
  • add further comments to object detection/tracking DNN example
  • add GetPerfProfile function to Net
  • add initial Layer implementation alongside enhancements to Net
  • add InputNameToIndex to Layer
  • add new functions allowing DNN backends such as OpenVINO
  • additional refactoring and comments in dnn-pose-detection example
  • cleanup DNN face detection example
  • correct const for device targets to be called Target
  • correct test that expected init slice with blank entries
  • do not init slice with blank entries, since added via append
  • further cleanup of DNN face detection example
  • make dnn-pose-detection example use Go channels for async operation
  • refactoring and additional comments for object detection/tracking DNN example
  • refine comment in header for style transfer example
  • working style transfer example
  • added ForwardLayers() to accomodate models with multiple output layers
• docs
  • add scripted Windows install info to README
  • Added a sample gocv workflow contributing guideline
  • mention docker image in README.
  • mention work in progress on Android
  • simplify and add missing step in Linux installation in README
  • update contributing instructions to match latest version
  • update ROADMAP from recent calib3d module contribution
  • update ROADMAP from recent imgproc histogram contribution
• examples
  • cleanup header for caffe dnn classifier
  • show how to use either Caffe or Tensorflow for DNN object detection
  • further improve dnn samples
  • rearrange and add comments to dnn style transfer example
  • remove old copy of pose detector
  • remove unused example
• features2d
  • free memory allocation bug for C.KeyPoints as pointed out by @tzununbekov
  • Adding opencv::drawKeypoints() support
• imgproc
  • add equalizeHist function
  • Added opencv::calcHist implementation
• openvino
  • add needed environment config to execute examples
  • further details in README explaining how to use
  • remove opencv contrib references as they are not included in OpenVINO
• videoio
  • Add OpenVideoCapture
  • Use gocv.VideoCaptureFile if string is specified for device.

v0.13.0

• build
  • Add cgo directives to contrib
  • contrib subpackage also needs cpp 11 or greater for a warning free build on Linux
  • Deprecate env scripts and update README
  • Don't set --std=c++1z on non-macOS
  • Remove CGO vars from CI and correct Windows cgo directives
  • Support pkg-config via cgo directives
  • we actually do need cpp 11 or greater for a warning free build on Linux
• docs
  • add a Github issue template to project
  • provide specific examples of using custom environment
• imgproc
  • add HoughLinesPWithParams() function
• openvino
  • add build tag specific to openvino
  • add roadmap info
  • add smoke test for ie

v0.12.0

• build
  • convert to CRLF
  • Enable verbosity for travisCI
  • Further improvements to Makefile
• core
  • Add Rotate, VConcat
  • Adding InScalarRange and NewMatFromScalarWithSize functions
  • Changed NewMatFromScalarWithSize to NewMatWithSizeFromScalar
  • implement CheckRange(), Determinant(), EigenNonSymmetric(), Min(), and MinMaxIdx() functions
  • implement PerspectiveTransform() and Sqrt() functions
  • implement Transform() and Transpose() functions
  • Make toByteArray safe for empty byte slices
  • Renamed InScalarRange to InRangeWithScalar
• docs
  • nicer error if we can't read haarcascade_frontalface_default
  • correct some ROADMAP links
  • Fix example command.
  • Fix executable name in help text.
  • update ROADMAP from recent contributions
• imgproc
  • add BoxFilter and SqBoxFilter functions
  • Fix the hack to convert C arrays to Go slices.
• videoio
  • Add isColor to VideoWriterFile
  • Check numerical parameters for gocv.VideoWriterFile
  • CodecString()
• features2d
  • add BFMatcher
• img_hash
  • Add contrib/img_hash module
  • add GoDocs for new img_hash module
  • Add img-similarity as an example for img_hash
• openvino
  • adds support for Intel OpenVINO toolkit PVL
  • starting experimental work on OpenVINO IE
  • update README files for Intel OpenVINO toolkit support
  • WIP on IE can load an IR network