20 years ago, in 1999, Kyocera launched the first mobile phone with a digital camera - Visual Phone VP-210. Since then, thanks to the incredibly large and growing market for mobile communication devices, CCD sensors of digital cameras have made an incredible leap in all respects. Sensitivity, range, size, power consumption, but more importantly, price.
In our realities, a camera module, actually a very technologically sophisticated device, can cost only a few dollars. This radically changes the view on many processes and tasks. Previously, the challenge was to get a camera that technically meets the minimum requirements. Having passed such a test, resolving image processing issues seemed only pleasant troubles. Now the issue of software that will process information from the camera is more acute. The bar of physical and economic access to technology has fallen so low that it has touched the boundaries of user competency.
Let's look at real-life examples of how difficult (or simple) it is now to work with images and what tasks are possible for an IT specialist of a different specialization.
Of course, the main tool for working with images is the Open Source OpenCV library. Written in C ++ - also has interfaces for working with Python, Java, PHP, JavaScript and other, less popular languages. On the example of several projects using OpenCV, published on the hub in 2018-2019, we will consider what tasks were solved and what technologies were used.
1) Smart feeder: Machine Learning, Raspberry Pi, Telegram, a little magic training + assembly instructions
Post on the hobby project ZlodeiBaal : 27.6k views, 289 bookmarks. Raspberry Pi B +, camera, OpenCV, Caffe, Python.
2) DIY thermal imager on Raspberry PI or “It seems now I know what I’ll do this summer”
A post on the Walker2000 hobby project: 73.8k views, 425 bookmarks. Raspberry Pi B + / Raspberry Pi Zero W, thermal imaging matrix, OpenCV, Python.
3) Print tapestry "Game of Thrones" on a fiscal printer using Python
Post about the hobby project viking_unet : 7.9k views, 50 bookmarks. Fiscal printer, OpenCV, Python.
4) StereoPi - our piece of hardware for studying computer vision, drones and robots
A post about the domestic hardware project Realizator : 14.1k views, 117 bookmarks. Raspberry Pi, OpenCV.
5) Restore blurry and defocused images using the Wiener filter. Implementation in C ++ OpenCV
Post about the experience of VladislavBK : 16.7k views, 154 bookmarks. Nikon, OpenCV, C ++.
6) OpenCV on STM32F7-Discovery
Post about the 0xdde hobby project: 6,4k views, 71 bookmarks. STM32, OpenCV, Qt, C ++.
7) Launch your neural network detector on the Raspberry Pi using the Neural Compute Stick and OpenVINO
Post about the hobby project BeloborodovDS : 14.7k views, 126 bookmarks. Raspberry Pi. Neural Compute Stick, OpenCV, OpenVINO, C ++.
8) Computer vision and machine learning in PHP using the opencv library
Post about the hobby project morozovsk : 21.6 thousand views, 236 bookmarks. OpenCV, PHP, php-opencv.
9) opencv4arts: Draw my city, Vincent
A post on computer vision and neural networks in the dkurt browser: 5.4k views, 63 bookmarks. OpenCV, JavaScript, OpenCV.js
10) * The place is vacant *
The last item was intended for a project in Java, since this language is officially officially supported by OpenCV, but I could not find a suitable post on Habré. Write your assumptions why? Better yet, write a post on the topic. Java is a very widespread language, and I am extremely surprised at such a meager representation on the hobby landscape of IT people.
It is not difficult to notice - in most cases, the authors first worked with OpenCV and computer vision in particular. This did not stop them with relatively small efforts to create a functioning project and even solve real problems in a convenient way.