Firefly high-performance face recognition SDK released

F_Developer_X

Firefly face recognition SDK released.

Now, we have entered the era of brush face. For example, face payment, face recognition access control, flow monitoring and so on. How to build DEMO in the Firefly open source board and achieve product quickly? In order to allow more products can be used face recognition technology, Firefly launched a high-performance face recognition SDK. The SDK's face recognition algorithm, which operates efficiently, has high recognition accuracy and excellent performance, it can be applied to commercial products directly.

For business applications, please contact us product@t-firefly.com

The following instructions are about how to deploy and test the SDK on Firefly RK3399 platform.

1. Install Ubuntu16.04 system firmware on Firefly RK3399 open source board

Download the system

https://drive.google.com/open?id=0B7HO8lbGgAqAX2oyZkg2QVQwZjA

Flash the firmware, refer here

http://wiki.t-firefly.com/index.php/Firefly-RK3399/Flash_image/en

2.  Download SDK

Download here

https://drive.google.com/open?id=0B7HO8lbGgAqAN1VqcW1OdWNQbDQ

Download and extract, put it into the Firefly RK3399 file system, such as ~ / directory

SDK directory description

include: SDK header files

   lib:  dynamic library files

  samples: The samples programs, FaceDetectImage is to detect faces in the picture, FaceDetectVideo is to detect faces in the video, FaceVerify is to compare faces in the picture.

3. Build operating environment

In the case of network connection, execute the script install_debs.sh which in the root directory of the SDK to a key install, the order is as follows:

./install_debs.sh dev

In the process, follow the prompts, you may need to enter the root password to get permission, or enter to confirm the installation

If the installation fails due to a network or other reason, try installing the following packages manually:

      sudo apt install libboost-all-dev
      sudo apt install libprotobuf-dev
      sudo apt install protobuf-compiler
      sudo apt install libopencv-dev
      sudo apt install libhdf5-dev
      sudo apt install libgflags-dev
      sudo apt install libgoogle-glog-dev
      sudo apt install libleveldb-dev
      sudo apt install liblmdb-dev
      sudo apt install libsnappy-dev

4. Run DEMO

Into the samples, in the corresponding directory to run the same name file, FaceDetectImage is to detect faces in the picture, FaceDetectVideo is to detect faces in the video, FaceVerify is to compare faces in the picture.

Note:  Run DEMO need root privileges, please add sudo to run DEMO

After FaceDetectImage run, it will detect faces in the specified image and add to the mark, as shown below:

After FaceDetectVideo run, it will read the frame images from the video file and add to the mark, as shown below:

Run FaceVerify will recognize the similarity of faces in two images and output a similarity value. The similarity value is between 0 and 1, the closer the value is 1, indicating the higher the similarity, vice versa. As shown below

5. More application scenarios

For example, based on the USB camera face recognition access control system, according to different people to provide different temperature settings of intelligent air conditioning and so on.

For business applications, please contact us product@t-firefly.com