newlandface

---

typora-copy-images-to: upload

newlandface

[TOC]

newlandface 是一个轻量级的人脸检测和多属性分析（年龄、性别、标签）分析工具，使用python语言，集成了部分开源库Dlib、mtcnn等工具，该库主要是基于Keras和TensorFlow来进行开发的。

Installation 安装

The easiest way to install newlandface is to download it from PyPI.

pip install newlandface

文件结构

│ README.md │ requirements.txt │ setup.py │
├─newlandface │ │ nlface.py │ │ init.py │ │
│ ├─basemodels │ │ │ DeepID.py │ │ │ DlibResNet.py │ │ │ Facenet.py │ │ │ FbDeepFace.py │ │ │ OpenFace.py │ │ │ VGGFace.py │ │ │ init.py │ │
│ ├─commons │ │ │ distance.py │ │ │ functions.py │ │ │ realtime.py │ │ │ init.py │ │
│ ├─extendedmodels │ │ │ Age.py │ │ │ Emotion.py │ │ │ Gender.py │ │ │ Race.py │ │ │ init.py │ │
│ ├─models │ │ face-recognition-ensemble-model.txt │ │ init.py │
└─tests │ testFaceAttr_video.py │ testFaceDetect_img.py │ testFacePoints_img.py │
└─dataset img1.jpg img13.jpg img14.jpg ...

测试代码

1.1 人脸检测代码

from newlandface import nlface
import cv2
# 模型加载
nlface.load_model()
image = cv2.imread("./dataset/test1.jpg")
# 人脸检测
faceObjs = nlface.detect_face(image)
# 显示人脸框
if faceObjs is not 0:
    for idx, rect in enumerate(faceObjs):
        image = nlface.show_face(image,rect)
else:
    print("no face detect")
    os._exit(0)
cv2.imshow("test",image)
cv2.waitKey()

1.2 人脸68点检测

1.2.1 直接调用show_face_points函数进行显示

核心函数：detect_face、show_face_points

from newlandface import nlface
import cv2
cv2.namedWindow("test",0)
# 模型加载
nlface.load_model()
image = cv2.imread("./dataset/test1.jpg")
# 人脸检测
faceObjs = nlface.detect_face(image)
# 显示人脸框
if faceObjs is not 0:
    for idx, rect in enumerate(faceObjs):
        image = nlface.show_face_points(image,rect)
else:
    print("no face detect")
    os._exit(0)
cv2.imshow("test",image)
cv2.waitKey()

1.2.2 调用点检测模块，自行画图

核心函数：detect_face、detect_points、show_face

from newlandface import nlface
import cv2
cv2.namedWindow("test",0)
# 模型加载
nlface.load_model()
image = cv2.imread("./dataset/test1.jpg")
# 人脸检测
faceObjs = nlface.detect_face(image)
if faceObjs is not 0:
    for idx, rect in enumerate(faceObjs):
        # 人脸68点检测
		points = nlface.detect_points(image,rect)
        # 显示人脸框、68点
		image = nlface.show_face(image,rect)
        for i,point in enumerate(points):
            cv2.circle(image,(point[0],point[1]),2,(0,0,255),-1)
            cv2.imshow("test",image)
        cv2.waitKey(1)    
else:
    print("no face detect")
    os._exit(0)
cv2.imshow("test",image)
cv2.waitKey()

1.3 人脸属性分析

核心函数：detect_face、detect_points、show_face

属性开放：emotion(表情)、age（年龄）、gender（性别）

属性	检测耗时	标签类型
emotion表情	30ms	['angry', 'disgust', 'fear', 'happy', 'sad', 'surprise', 'neutral']
age年龄	130ms	1-100
gender性别	170ms	woman、man

注意：不同的模块耗时不一，所以如果调用摄像头的时候，要注意实时性上的要求。

from newlandface import nlface
import cv2
cv2.namedWindow("test",0)
# 模型加载
nlface.load_model()
image = cv2.imread("./dataset/test1.jpg")
# 人脸检测
faceObjs = nlface.detect_face(image)
if faceObjs is not 0:
    for idx, rect in enumerate(faceObjs):
        # 人脸属性分析
        actions = ['emotion', 'age', 'gender']
        attribute = nlface.analyze(image, faceObjs[idex],actions = actions)
        # 显示人脸框\属性
		image = nlface.show_face(image,rect)
        image = nlface.show_face_attr(image, faceObjs[idex], attribute, actions)
        cv2.imshow("test",image)
        cv2.waitKey(1)    
else:
    print("no face detect")
    os._exit(0)
cv2.imshow("test",image)
cv2.waitKey()