Skip to content

evaluation of face detectors including opencv dnn face detector

License

Notifications You must be signed in to change notification settings

KatsunoriWa/eval_faceDetectors

Repository files navigation

eval_faceDetectors

evaluation of face detectors including opencv dnn face detector

  • You can evaluate face detection in each face detector in Jupyter environment.

  • Some modules are wrapped by class interface

Haar Cascade Face Detector in opencv

resnet ssd Face Detector in opencv

dlib frontal Face detector

dlib CNN Face detector

tensorflow Face detector

detector pretrained_model input image size
OpenCV Haar Cascade haarcascade_frontalface_default.xml no upper limit
OpenCV resnet ssd face detcator res10_300x300_ssd_iter_140000.caffemodel 300x300
dlib FrontalFace
dlib cnn_face_detector mmod_human_face_detector.dat
tensorflow Face detector frozen_inference_graph_face.pb

memo in Japanese

顔検出器の評価のポイント

  • 検出率
  • 検出にかかる時間・計算量

顔画像の検出器は、撮影される画像の素性による。 LFWの画像の場合は、 顔照合を目的とした画像セットであるため、 顔の顔向きがほぼ正面顔である。

About

evaluation of face detectors including opencv dnn face detector

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published