OmarSayedMostafa / All-in-one-FaceDetection

this is a collected repos for most famous algorithms and models for face detection.

All-in-one-FaceDetection

this is a collected repos for most famous algorithms and models for face detection collocted from github repos and structed in more organised way for speed and accuracy comparison.

• each enviroment specify which model would work best for it from speed and accuracy prespective

this repos contains:

• Cloned from learnopencv website // github Highly recommend to visit the website so you will know more about how the algorithms/model work and for further more understanding, also it contains a wide material and tutorial on Computer Vision tasks.

  • OpenCV

    • Opencv classisc haarcascade.
    • Opencv DNN
      • Single-Shot-Multibox detector SSD and uses ResNet-10 Architecture as backbone.
        1. FloatPoint16 version of the original caffe implementation ( 5.4 MB )
      2. 8 bit Quantized version using Tensorflow ( 2.7 MB )

  • Dlib

    1. Dlib MMOD DNN : Maximum-Margin Object Detector ( MMOD ) with CNN based features.
    2. Dlib HOG Detector: Based on HoG features and SVM

• MTCNN Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks

• face_recognition orginal repo Built using dlib's state-of-the-art face recognition built with deep learning. The model has an accuracy of 99.38% on the Labeled Faces in the Wild benchmark.

• YoloV3 From DarkNet cloned from these github repo and github repo

• UltraLightOnnxFaceDetector by Linzaer and MobileFaceNet,

  • the most accurate and fast worked for me on cpu, cloned from Medium blog post // Github Repo by [Author: fyr91]