OpenCV (Open Source Computer Vision Library) is the world’s largest open-source computer vision library and a foundational dependency in essentially every commercial robot with a camera. Originally an Intel Research Labs project that began in 1999 and was first publicly released in 2000, it is today maintained by the non-profit Open Source Vision Foundation and has been licensed under Apache 2 since version 4.5. The library contains over 2,500 optimised algorithms spanning image processing, feature detection and description (SIFT, ORB, AKAZE), camera calibration and stereo vision, optical flow, object detection, segmentation, structure-from-motion, and a complete general-purpose machine learning module — plus integrations with deep-learning runtimes via the dnn module supporting models from TensorFlow, PyTorch, Caffe, and ONNX. The codebase is cross-platform across Windows, Linux, macOS, Android, and iOS, with C++, Python, Java, and JavaScript bindings. CUDA, OpenCL, and Apple-specific GPU acceleration are available for real-time work. ROS uses OpenCV as its main vision package, and the library underpins everything from drone autopilots and surgical robots to factory inspection systems and humanoid perception stacks. Citing OpenCV remains essentially mandatory for any computer-vision research paper.
The de-facto open-source computer vision library: 2,500+ optimised algorithms covering image processing, feature detection, calibration, and classical and deep-learning vision. Cross-platform with C++, Python, Java, and JavaScript bindings. Foundational dependency in essentially every robotics vision stack.
OpenCV (Open Source Computer Vision Library) is the world’s largest open-source computer vision library and a foundational dependency in essentially every commercial robot with a camera. Originally an Intel Research Labs project that began in 1999 and was first publicly released in 2000, it is today maintained by the non-profit Open Source Vision Foundation and has been licensed under Apache 2 since version 4.5. The library contains over 2,500 optimised algorithms spanning image processing, feature detection and description (SIFT, ORB, AKAZE), camera calibration and stereo vision, optical flow, object detection, segmentation, structure-from-motion, and a complete general-purpose machine learning module — plus integrations with deep-learning runtimes via the dnn module supporting models from TensorFlow, PyTorch, Caffe, and ONNX. The codebase is cross-platform across Windows, Linux, macOS, Android, and iOS, with C++, Python, Java, and JavaScript bindings. CUDA, OpenCL, and Apple-specific GPU acceleration are available for real-time work. ROS uses OpenCV as its main vision package, and the library underpins everything from drone autopilots and surgical robots to factory inspection systems and humanoid perception stacks. Citing OpenCV remains essentially mandatory for any computer-vision research paper.