Installer opencv avec Homebrew, apk, dnf, Nix, pacman, zypper, chocolatey, apt, scoop

Liens du projet

• Page d'accueil
• Dépôt
• Docs amont
• Page du gestionnaire de paquets

Commandes et alias

• opencv_annotation
• opencv_interactive-calibration
• opencv_model_diagnostics
• opencv_version
• opencv_visualisation
• opencv_waldboost_detector
• setup_vars_opencv4.sh

Historique du projet

OpenCV began inside Intel. The OpenCV anniversary timeline says Gary Bradski proposed a computer-vision library in 1998, Intel open-sourced the code in 1999, and the first public release was unveiled at CVPR in June 2000. The project name followed the OpenGL naming style and reflected a shift from an internal Computer Vision Library concept to an open source library.

Version 1.0 arrived in 2006 with a C API and classic computer-vision capabilities such as image processing, computational geometry, face detection, camera calibration, optical flow, SIFT features, and machine-learning methods. After 2008, development moved through Willow Garage and Itseez, connecting OpenCV to robotics and specialized vision-algorithm work.

OpenCV 2.0 in 2009-2010 made C++ the primary API and added generated Python bindings, with Java bindings added as well. In 2012 the project moved to GitHub and opencv.org, creating a more transparent contribution process; the anniversary timeline says 35 to 50 percent of accepted pull requests in 2013 came from outside the core team.

Intel acquired Itseez in 2016, bringing a core OpenCV development team back into Intel's orbit. OpenCV 3.0, OpenCV 4.0, the DNN module, JavaScript/WebAssembly work, and OpenVINO integration all reflect the library's shift from classic computer vision only toward heterogeneous, deep-learning-aware, edge and embedded deployment.

Historique d'adoption

OpenCV's own About page describes it as the world's biggest computer vision library, with more than 2500 algorithms, hundreds of thousands of users, and estimated monthly downloads exceeding 40 million. The same page names use by companies, research groups, hobbyists, government bodies such as NASA, and companies including Google, Yahoo, Microsoft, Intel, IBM, Sony, Honda, and Toyota.

The project spread because it gave computer-vision researchers and product developers a permissively licensed, optimized common infrastructure. Instead of each application reimplementing camera calibration, feature detection, object detection, video I/O, image filtering, machine learning, and matrix primitives, developers could depend on a common package that existed in Linux distributions, Homebrew, Chocolatey, Conda-like scientific stacks, mobile builds, and language bindings.

Modes d'utilisation

Developers use OpenCV for frame capture, image preprocessing, camera calibration, feature extraction, object detection, tracking, stereo reconstruction, panorama stitching, face/eye detection, DNN inference, and high-level GUI/debug display. The modern documentation describes a modular structure with core, imgproc, imgcodecs, videoio, highgui, video, calib3d, features2d, objdetect, dnn, ml, photo, stitching, and supporting modules.

In package-manager form, OpenCV is often consumed as a library dependency rather than as an end-user CLI. The executables such as `opencv_version`, annotation tools, calibration tools, visualization helpers, and setup scripts are useful, but most downstream value comes from headers, shared libraries, Python bindings, and integrations into robotics, vision, scientific, and AI applications.

Pourquoi les passionnés de paquets s'y intéressent

OpenCV is package-nerd significant because it is large, modular, ABI-sensitive, and option-heavy. Builds may vary by language bindings, codecs, GUI backends, CUDA/OpenCL/OpenVINO acceleration, contrib modules, platform SIMD, and patent-sensitive algorithm history, so the same package name can imply very different capabilities across ecosystems.

It is also a dependency magnet: robotics frameworks, scientific tools, media applications, embedded vision products, and AI demos all pull it in. That makes versioning, binary size, transitive multimedia dependencies, and Python/C++ ABI compatibility recurring concerns for maintainers.

Chronologie

• 1998: Gary Bradski proposed the library inside Intel.
• 1999: Intel open-sourced the library and the OpenCV name replaced the internal CVL direction.
• 2000-06: The first public release was unveiled at CVPR.
• 2006-10: OpenCV 1.0 was released.
• 2008: Core development moved through Willow Garage and Itseez.
• 2009-2010: OpenCV 2.0 made C++ the primary API and expanded generated language bindings.
• 2012: The project moved to opencv.org and GitHub.
• 2015: OpenCV 3.0 added major interface and acceleration improvements.
• 2016: Intel acquired Itseez, a key contributor to OpenCV and OpenVX.
• 2018: OpenCV 4.0 refined the API around newer C++ standards and added Graph API work.

Related projects

• OpenCV is related to OpenVINO, OpenVX, ROS and Willow Garage-era robotics software, opencv_contrib, NumPy/Python scientific tooling, CUDA/OpenCL acceleration stacks, and downstream computer-vision frameworks that use it as preprocessing or classical-vision infrastructure.

Sources

• https://docs.opencv.org/4.x/d1/dfb/intro.html
• https://github.com/opencv/opencv
• https://opencv.org/about/
• https://opencv.org/anniversary/
• https://opencv.org/blog/intel-acquires-itseez/