A brief history of the project OpenCV
Author: Kirill Kornyakov, leader of the development team in Itseez.
Hi, Habr!
')
With this post we open the Itseez blog (from the English sees it - “it sees!”). The company was founded by adherents of computer vision, who have been working in this field for more than 13 years, among which there are participants of the OpenCV project from the first day of its existence. Our strategy is connected with the development of technologies and products in the field of computer vision under the motto “Vision that works!”. Our clients are Willow Garage , NVidia, Intel, Microsoft, several lesser-known companies and start-ups. Today, the Itseez team is a team of 40 engineers and researchers, many employees actively participate in educational activities: they teach at the university, publish in journals and speak at conferences, and also conduct schools on computer vision.
On the blog pages, we would like to talk about what is happening in the real and cyber space around us, as well as in the world of computer vision. We hope that the published materials will be of interest to people close to machine learning, robotics, computational photography, intelligent video surveillance technologies and augmented reality, as well as to related areas of Science and Technology. An important feature of Itseez is that many of our developments are open-source, so we will try to back up our words with examples of real code.
Since this is the first entry in our blog, I would like to share something interesting, so we will talk about the company's key project - the development of the library of computer vision algorithms OpenCV (about our contributions to the ROS (Robot Operating System) and PCL (Point Cloud Library) we keep silent for now). The prepared reader will probably be surprised at the question: “But after all, OpenCV is being developed by Intel / Willow Garage!”, And it will be right, but only partially. Indeed, not everyone knows that most of the leading developers of OpenCV live and work in Russia, in the city of Nizhny Novgorod, and are employees of the Itseez company. Therefore, in order to eliminate misunderstandings, we decided to devote the first post to a brief summary of the history of OpenCV. Also, taking this opportunity, in conclusion we will share some insider information about the future of the project.
So, if you do computer vision, then you probably heard about OpenCV, the full name of which sounds like the Open Source Computer Vision Library. The library is quite popular, currently has just over 5 million downloads (not counting downloads directly from the repository), and was recently proposed as the basis for the Khronos standard for computer vision ( http://www.khronos.org/vision ).
From 2000 to 2008, OpenCV was developed and supported primarily by Intel, and the Nizhny Novgorod branch of the corporation from the very beginning played a leading role in shaping the image of the library. In the early years of its existence, OpenCV rapidly expanded in breadth, acquiring basic functionality, such as basic data structures, image processing algorithms (image processing), basic computer vision algorithms, image and video input and output. Already at that time, algorithms for detecting human faces (cascade classifier), stereo matching, optical flow, and others were implemented. However, in 2004, Intel virtually ceased supporting the library. Many of the project participants, who by that time had already formed as experts in the field of computer vision, did not want to retrain, and left the company. In fact, this is how Itseez originated, whose activities are largely based on the experience gained during the development of OpenCV.
The second powerful impetus to the development of the project received with the arrival of the company Willow Garage, whose main goal is to create a personal robot. OpenCV has become an important part of the ROS (Robot Operating System), and on its basis a number of components were created for the PR2 robot, but we will tell about this sometime later. The beginning of this period coincides with the transition to Itseez from Intel by Vadim Pisarevsky (aka OpenCV Tsar) in 2008. With the support of Willow, a team was formed that launched a substantial library rework. It is the result of these efforts that OpenCV acquired C ++ and Python API, features2d module, new architecture, CMake-based build system, BuildBot-based continuous integration system, improved documentation, excellent tutorials and a host of other innovations.
Today we can talk about the third significant stage in the library’s life, which came with the advent of NVidia. In 2010, this company supported the creation of a CUDA-optimized version of the library, which was also done by Itseez engineers. The first public result was the implementation of a stereo-matching algorithm, capable of real-time video processing in FullHD-resolution (1920 × 1080 pixels). Today, opencv_gpu is a full-fledged module that has found application in many industrial applications. Since 2011, the task list has expanded, and Itseez, together with NVidia, began developing a version of the library for Android, with optimizations for the NVidia Tegra platform. OpenCV4Android has already experienced several releases, and has acquired its own user community . In addition, OpenCV has been enriched by the Java API (currently only for Android), the built-in performance testing system and a number of improvements. Those who may be interested in the results of OpenCV optimization for GPUs and mobile processors can be recommended to read the recently published article “Real-time Computer Vision with OpenCV” , as well as posters from the GTC 2012 conference: “Accelerating Computer Vision with Tegra GPU” , “GPU Vision .
Thus, the modern look of the library is the result of a long evolution. The project keeps pace with the rapidly developing field of computer vision. Several world-class companies participated in the formation of the library, and it is still too early to talk about the completion of its development. Recently, the project has acquired a new tracker of tasks, activity is only gaining momentum, an anniversary 5M download recently took place, and we hope that we will have a lot of interesting material to share with you.
Now some information for active users of OpenCV. First, on June 4, the release 2.4.1 was released, mainly carrying corrections for 2.4.0 (special thanks to Andrei Kamaev and Alexander Shishkov!). Secondly, in the near future we plan to launch our own library website, which will be located at opencv.org (now the redirect is working). This site will be especially useful to those who are just starting to work with the library. In addition, it will deploy its own Q & A forum on the StackOverflow-like engine. Thirdly, work is underway to create the OpenCV Foundation, a non-profit foundation that would allow raising funds for the development of the library and systematizing its development. For developers, this means even more functionality, documentation, samples and performance! We really hope that all this will happen this year, we will try to keep everyone informed!
Finally, we want to invite everyone to familiarize themselves with the visualization of activity in the repository while working on OpenCV 2.4.0 (see the maximum resolution!).
The next major release of OpenCV 2.5 is scheduled for November 1, and work is well underway!
Hi, Habr!
')
With this post we open the Itseez blog (from the English sees it - “it sees!”). The company was founded by adherents of computer vision, who have been working in this field for more than 13 years, among which there are participants of the OpenCV project from the first day of its existence. Our strategy is connected with the development of technologies and products in the field of computer vision under the motto “Vision that works!”. Our clients are Willow Garage , NVidia, Intel, Microsoft, several lesser-known companies and start-ups. Today, the Itseez team is a team of 40 engineers and researchers, many employees actively participate in educational activities: they teach at the university, publish in journals and speak at conferences, and also conduct schools on computer vision.
On the blog pages, we would like to talk about what is happening in the real and cyber space around us, as well as in the world of computer vision. We hope that the published materials will be of interest to people close to machine learning, robotics, computational photography, intelligent video surveillance technologies and augmented reality, as well as to related areas of Science and Technology. An important feature of Itseez is that many of our developments are open-source, so we will try to back up our words with examples of real code.
Since this is the first entry in our blog, I would like to share something interesting, so we will talk about the company's key project - the development of the library of computer vision algorithms OpenCV (about our contributions to the ROS (Robot Operating System) and PCL (Point Cloud Library) we keep silent for now). The prepared reader will probably be surprised at the question: “But after all, OpenCV is being developed by Intel / Willow Garage!”, And it will be right, but only partially. Indeed, not everyone knows that most of the leading developers of OpenCV live and work in Russia, in the city of Nizhny Novgorod, and are employees of the Itseez company. Therefore, in order to eliminate misunderstandings, we decided to devote the first post to a brief summary of the history of OpenCV. Also, taking this opportunity, in conclusion we will share some insider information about the future of the project.
So, if you do computer vision, then you probably heard about OpenCV, the full name of which sounds like the Open Source Computer Vision Library. The library is quite popular, currently has just over 5 million downloads (not counting downloads directly from the repository), and was recently proposed as the basis for the Khronos standard for computer vision ( http://www.khronos.org/vision ).
From 2000 to 2008, OpenCV was developed and supported primarily by Intel, and the Nizhny Novgorod branch of the corporation from the very beginning played a leading role in shaping the image of the library. In the early years of its existence, OpenCV rapidly expanded in breadth, acquiring basic functionality, such as basic data structures, image processing algorithms (image processing), basic computer vision algorithms, image and video input and output. Already at that time, algorithms for detecting human faces (cascade classifier), stereo matching, optical flow, and others were implemented. However, in 2004, Intel virtually ceased supporting the library. Many of the project participants, who by that time had already formed as experts in the field of computer vision, did not want to retrain, and left the company. In fact, this is how Itseez originated, whose activities are largely based on the experience gained during the development of OpenCV.
The second powerful impetus to the development of the project received with the arrival of the company Willow Garage, whose main goal is to create a personal robot. OpenCV has become an important part of the ROS (Robot Operating System), and on its basis a number of components were created for the PR2 robot, but we will tell about this sometime later. The beginning of this period coincides with the transition to Itseez from Intel by Vadim Pisarevsky (aka OpenCV Tsar) in 2008. With the support of Willow, a team was formed that launched a substantial library rework. It is the result of these efforts that OpenCV acquired C ++ and Python API, features2d module, new architecture, CMake-based build system, BuildBot-based continuous integration system, improved documentation, excellent tutorials and a host of other innovations.
Today we can talk about the third significant stage in the library’s life, which came with the advent of NVidia. In 2010, this company supported the creation of a CUDA-optimized version of the library, which was also done by Itseez engineers. The first public result was the implementation of a stereo-matching algorithm, capable of real-time video processing in FullHD-resolution (1920 × 1080 pixels). Today, opencv_gpu is a full-fledged module that has found application in many industrial applications. Since 2011, the task list has expanded, and Itseez, together with NVidia, began developing a version of the library for Android, with optimizations for the NVidia Tegra platform. OpenCV4Android has already experienced several releases, and has acquired its own user community . In addition, OpenCV has been enriched by the Java API (currently only for Android), the built-in performance testing system and a number of improvements. Those who may be interested in the results of OpenCV optimization for GPUs and mobile processors can be recommended to read the recently published article “Real-time Computer Vision with OpenCV” , as well as posters from the GTC 2012 conference: “Accelerating Computer Vision with Tegra GPU” , “GPU Vision .
Thus, the modern look of the library is the result of a long evolution. The project keeps pace with the rapidly developing field of computer vision. Several world-class companies participated in the formation of the library, and it is still too early to talk about the completion of its development. Recently, the project has acquired a new tracker of tasks, activity is only gaining momentum, an anniversary 5M download recently took place, and we hope that we will have a lot of interesting material to share with you.
Now some information for active users of OpenCV. First, on June 4, the release 2.4.1 was released, mainly carrying corrections for 2.4.0 (special thanks to Andrei Kamaev and Alexander Shishkov!). Secondly, in the near future we plan to launch our own library website, which will be located at opencv.org (now the redirect is working). This site will be especially useful to those who are just starting to work with the library. In addition, it will deploy its own Q & A forum on the StackOverflow-like engine. Thirdly, work is underway to create the OpenCV Foundation, a non-profit foundation that would allow raising funds for the development of the library and systematizing its development. For developers, this means even more functionality, documentation, samples and performance! We really hope that all this will happen this year, we will try to keep everyone informed!
Finally, we want to invite everyone to familiarize themselves with the visualization of activity in the repository while working on OpenCV 2.4.0 (see the maximum resolution!).
The next major release of OpenCV 2.5 is scheduled for November 1, and work is well underway!
Source: https://habr.com/ru/post/146434/
All Articles