A Golang binding for OpenCV.

OpenCV 1.x C API bindings through CGO, and OpenCV 2+ C++ API (GoCV) through SWIG.

DISCLAIMER

Install Go and OpenCV, you might want to install both of them via apt-get or homebrew.

go get github.com/lazywei/go-opencv
cd $GOPATH/src/github.com/lazywei/go-opencv/samples
go run hellocv.go

• Install Go and MinGw
• install OpenCV-2.4.x to MinGW dir

# libopencv*.dll --> ${MinGWRoot}\bin
# libopencv*.lib --> ${MinGWRoot}\lib
# include\opencv --> ${MinGWRoot}\include\opencv
# include\opencv2 --> ${MinGWRoot}\include\opencv2

go get github.com/lazywei/go-opencv
cd ${GoOpenCVRoot}/trunk/samples && go run hellocv.go

After OpenCV 2.x+, the core team no longer develop and maintain C API. Therefore, CGO will not be used in CV2 binding. Instead, we are using SWIG for wrapping. The support for OpenCV2 is currently under development, and whole code will be placed under gocv package.

If you want to use CV2's API, please refer to the code under gocv/ directory. There is no too many documents for CV2 wrapper yet, but you can still find the example usages in *_test.go.

Please also note that the basic data structures in OpenCV (e.g., cv::Mat, cv::Point3f) are wrapped partially for now. For more detail on how to use these types, please refer to GoCV's README.

Requirement: we will build the wrappers based on mat64, given it is much easier to manipulate the underlaying data. In most case, it is not necessary to access the original CV data, e.g., cv::Mat can be converted from/to *mat64.Dense.

package main

import . "github.com/lazywei/go-opencv/gocv"
import "github.com/gonum/matrix/mat64"

func main() {

	objPts := mat64.NewDense(4, 3, []float64{
		0, 25, 0,
		0, -25, 0,
		-47, 25, 0,
		-47, -25, 0})

	imgPts := mat64.NewDense(4, 2, []float64{
		1136.4140625, 1041.89208984,
		1845.33190918, 671.39581299,
		302.73373413, 634.79998779,
		1051.46154785, 352.76107788})

	camMat := GcvInitCameraMatrix2D(objPts, imgPts)
	fmt.Println(camMat)
}

package main

import opencv "github.com/lazywei/go-opencv/opencv"

func main() {
	filename := "bert.jpg"
	srcImg := opencv.LoadImage(filename)
	if srcImg == nil {
		panic("Loading Image failed")
	}
	defer srcImg.Release()
	resized1 := opencv.Resize(srcImg, 400, 0, 0)
	resized2 := opencv.Resize(srcImg, 300, 500, 0)
	resized3 := opencv.Resize(srcImg, 300, 500, 2)
	opencv.SaveImage("resized1.jpg", resized1, 0)
	opencv.SaveImage("resized2.jpg", resized2, 0)
	opencv.SaveImage("resized3.jpg", resized3, 0)
}

Yet another cool example is created by @saratovsource which demos how to use webcam:

cd samples
go run webcam.go

You can find more samples at: https://github.com/lazywei/go-opencv/tree/master/samples

• Fork this repo

• Clone the main repo, and add your fork as a remote

  git clone https://github.com/lazywei/go-opencv.git
  cd go-opencv
  git remote rename origin upstream
  git remote add origin https://github.com/your_github_account/go-opencv.git
  

• Create new feature branch

  git checkout -b your-feature-branch
  

• Commit your change and push it to your repo

  git commit -m 'new feature'
  git push origin your-feature-branch
  

• Open a pull request!

This is a fork of chai's go-opencv, which has only OpenCV1 support through CGO. At the time of the fork (Dec 9, 2013) the original project was inactive, and hence I decide to host a fork on Github so people can contribute to this project easily. However, now it seems to be active again starting from Aug 25, 2014. Efforts to merge the two projects are very welcome.