AR, beauty, Robot: Computer Vision Library almost ubiquitous

The recent launch of the anti-beauty app Primo in Japan may make you feel overwhelmed. In fact, this anti-human application, you can also write, but must understand some of the technology, is computer vision. At present, the computer Vision Library includes FASTCV, OpenCV, JAVACV and so on.

Relatively speaking, OpenCV is a more mature visual library, it contains the Harris, SURF, SIFT, fast and other algorithms, support the object-oriented C + + API, and can be optimized for different hardware, such as desktops, mobile devices and so on.

JAVACV is a library that encapsulates common library interfaces for computer visual programmers such as OpenCV, libdc1394, Openkinect, Videoinput, and Artoolkitplus. If you are developing an app that doesn't need to add code to work with pictures, you can use JAVACV. But if Uoxuyao a lot of extra code for image processing, then Java will slow down your processing.

The FASTCV is a computer vision library optimized for mobile devices. The Qualcomm Augmented Reality (AR) SDK is a good example of how developers can use the framework required by FASTCV to build computer vision applications. FASTCV can add more camera-based features to developers ' applications, such as gesture recognition, text recognition, augmented reality, and face detection, tracking, and recognition.

Qualcomm's Snapdragon series (S2 version above) is a mobile device processor developed based on the ARM architecture. Accordingly, the FASTCV supports all ARM processors and is optimized for the Qualcomm processor.

To cite an example, this year Qualcomm launched the Dragon 805 series chip. The Android phone with this chip can realize the function of "take photos first, then focus", that is to say, users can set the focus of the photo by clicking on objects of different distance in the photo, which makes use of FASTCV.

Ionroad Application screenshot

The Ionroad is a mobile phone app that can take advantage of mobile phone cameras and built-in GPS, accelerometers, and gyroscopes to get information about the speed and direction of your vehicle, and it can also detect front vehicles and lanes through camera monitoring, and alert you when you're off the road or close to the car. This is the ability to use the FASTCV to monitor and track objects. Although the app was developed by the developer in 2012 when FASTCV was just released. However, according to the official introduction, the use of FASTCV, performance improved by 10%-15%, two days to complete the development.

In fact, computer vision can not only be used in photography, AR, or other camera-related mobile applications, for the new field of robotics is also very important. For example, at this year's UPLINQ conference, developers exhibited a robot product that used the Snapdragon 600 processor to support the FASTCV Visual Computing library and use the camera to identify and track objects.

To learn more about FASTCV and Qualcomm Technical information, please visit the Qualcomm Developer Zone

AR, beauty, Robot: Computer Vision Library almost ubiquitous