Skip to content

Latest commit

 

History

History
128 lines (110 loc) · 3.74 KB

README.md

File metadata and controls

128 lines (110 loc) · 3.74 KB
Raw

face_recognize

Build face recognition application easily. This library provides a simple way to use dlib's state-of-the-art face recognition and ArcSoft's free offline face recognition SDK (which is quicker and more robust than dlib model).

Installation

pip install face_recognize

Installation Options:

This library supports three backends for now:

  • arcsoft_v1: ArcSoft Face Recognition SDK 1.0
  • arcsoft_v3: ArcSoft Face Recognition SDK 2.0/3.0
  • dlib
Enable dlib backend
  • pip install dlib
  • pip install face_recognition_models
Enable arcsoft_v1
  • Download ArcSoft Face Recognition SDK 1.0 from ArcSoft.
  • face_recognize init
  • Move libarcsoft_fsdk_face_detection.dll(.so), libarcsoft_fsdk_face_recognition.dll(.so), libarcsoft_fsdk_face_tracking.dll(.so) to $HOME/.face_recognize/lib.
  • Modify $HOME/.face_recognize/config/arcsoft_v1_config.py and set the APPID, KEY.
Enable arcsoft_v3
  • Download ArcSoft Face Recognition SDK 3.0 from ArcSoft.
  • face_recognize init
  • Move libarcsoft_face.dll(.so), libarcsoft_face_engine.dll(.so) to $HOME/.face_recognize/lib.
  • Modify $HOME/.face_recognize/config/arcsoft_v3_config.py and set the APPID, KEY.

Usage

CLI

Register user with image

face_recognize register --image path-to-image (--name user-name)

Recognize people in an image

face_recognize recognize --image path-to-image

Recognize people in a video

face_recognize recognize --video path-to-video

Recognize people in usb camera

face_recognize recognize --video 0

Delete User Feature in Database

face_recognzie delete --name user-name

Delete All Features in Database

face_recognzie clear

The backend is default to be arcsoft_v3, use --verion backend-verson to indicate the backend.

Python Module

Face Detection

from face_recognize import FaceDetector
import cv2
img = cv2.imread('sample.jpg')
detector = FaceDetector()
infos = detector.detect(img)
detector.drawInfos(img, infos, show_name=False)
cv2.imshow('img', img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Face Feature Extraction

from face_recognize import FaceDetector, FaceRecognizer
import cv2
img = cv2.imread('sample.jpg')
detector = FaceDetector()
recognizer = FaceRecognizer()
infos = detector.detect(img)
feature = recognizer.extract(img, infos[0])

Face Feature Compare

from face_recognize import FaceDetector, FaceRecognizer
import cv2
img1 = cv2.imread('me.jpg')
img2 = cv2.imread('unknown.jpg')
detector = FaceDetector()
recognizer = FaceRecognizer()
infos1 = detector.detect(img1)
infos2 = detector.detect(img2)
feature1 = recognizer.extract(img1, infos1[0])
feature2 = recognizer.extract(img2, infos2[0])

if recognizer.judge(feature1, feature2):
    print("This is me")
else:
    print("This is not me")

Face Recognition (without db)

from face_recognize import FaceDetector, FaceRecognizer
import cv2

detector = FaceDetector()
recognizer = FaceRecognizer()
recognizer.register_feature(['me.jpg'], name=['me'], to_db=False, to_buffer=True, detector=detector)

img = cv2.imread('unknown.jpg')
infos = detector.detect(img)
feature = recognizer.extract(img, infos[0])
name = recognizer.recognize(feature)
if name:
    print("This is %s" % name)
else:
    print("Unknown")

Recognize People in Usb Camera (with db)

from face_recognize import FaceDetector
from cv2_utils import VideoCapture
import cv2

detector = FaceDetector("arcsoft_v3")

cap = VideoCapture(0, show_video=False)
for img in cap:
    face_infos = detector.track(img)
    img = detector.drawInfos(img, face_infos)
    cv2.imshow('img', img)