Brodmann17 Face Detector

You can use this forked repo to detect face in image/images/webcam and also for benchmarking face detection performance in AFW,PASCAL,FDDB,UFDD and WIDER face dataset.

From Authors: We share here a version of Brodmann17's super fast and accurate face detection library.
This detector is based on a proprietary deep-learning algorithm which allows to run highly accurate CNNs
efficiently on ARM CPUs.

Disclaimer:

• This library is shared for non-commercial use only. For advanced, more accurate version for commercial use please reach out at [email protected]

Library Dependencies

• OpenBLAS 0.2.18 Sourcecode

• google protobuf 2.6.1 Sourcecode

  If you are using ubuntu 16.04 these libraries can installed by:

    sudo apt-get install libopenblas-dev libprotobuf-dev

Supported Architectures

• Armv8a (aarch64) : linux, tested on Ubuntu 16.04
• Armv7l (arm) : linux, tested on Fedora 22
• Intel (x84_64) : linux, tested on Ubuntu 16.04

Speed Benchmark (1 x ARM cortex A72, Rockchip RK3399)

Input Image Size	Process Time [ms]	FPS [1/s]
640x480	67.72	14.77
320x240	22.26	44.93

---

Getting started (C++)

• Requirements:
• CMake (at least 3.0.0) - can be installed by using:

        sudo apt-get install cmake

• OpenCV (tested with 2.4.9.1) - can be installed by using:

        sudo apt-get install libopencv-dev

• Run these commands

        cd example 
        mkdir build  
        cd build  
        cmake .. 
        make
        cd ..

1. Test webcam

$./build/example -mode=0 -webcam=0

2. Test single image

$./build/example -mode=1 -path=../image/1.jpg

3. Test image lists

$./build/example -mode=2 -path=../image/

4. Evaluation in benchmark dataset, detection files will be stored in example/detections folder.

a) afw dataset
$./build/example -mode=3 -dataset=AFW -path=/path/to/afw/dataset/

b) PASCAL dataset
$./build/example -mode=3 -dataset=PASCAL -path=/path/to/pascal/dataset/

c) FDDB dataset
$./build/example -mode=3 -dataset=FDDB -path=/path/to/fddb/dataset/

d) WIDER_val dataset
#./build/example -mode=3 -dataset=WIDER_VAL -path=/path/to/wider/validation/dataset/

e) UFDD dataset
#./build/example -mode=3 -dataset=UFDD -path=/path/to/UFDD/validation/dataset/

Getting started (python)

• Requirements:
  • Python 2.x
  • Numpy
  • Matplotlib
  • OpenCV

        pip install numpy, matplotlib, opencv-python

• Usage

        from brodmann17_face_detector import Detector
    
        im = cv2.imread("../example/example2.jpg")
        with Detector() as det:
            detections = det.detect(im)

See python/example.ipynb for a full working example