2020-01-09：InsightFace项目实战（四）调用模型检测_insightface模型输出

一、参考资料

https://zhuanlan.zhihu.com/p/43804018

二、实战

1、cd到如下目录：insightface/deploy/test.py，test.py的input是图片，output是人脸特征

2、复制test.py，并修改文件名为test_mines.py

3、修改test_mines.py

• 修改模型存储路径：r100模型存储路径和性别年龄模型存储路径
• 修改数据存储路径
• 用余弦距离计算相似度，与设定阈值比较，输出检测结果

4、test_mines.py代码如下

import face_model_mines
import argparse
import cv2
import sys
import numpy as np
 
parser = argparse.ArgumentParser(description='face model test')
# general
parser.add_argument('--image-size', default='112,112', help='')
parser.add_argument('--model', default='/data/tensorflow/WorkSpace/syd/FaceRecognition/Projects/insightface/recognition/models/model-r100-ii/r100-arcface-emore/model,0555', help='path to load model.')
parser.add_argument('--ga-model', default='/data/tensorflow/WorkSpace/syd/FaceRecognition/Projects/insightface/models/gamodel-r50/model,0', help='path to load model.')
parser.add_argument('--gpu', default=0, type=int, help='gpu id')
parser.add_argument('--det', default=0, type=int, help='mtcnn option, 1 means using R+O, 0 means detect from begining')
parser.add_argument('--flip', default=0, type=int, help='whether do lr flip aug')
parser.add_argument('--threshold', default=1.24, type=float, help='ver dist threshold')
args = parser.parse_args()
 
##模型载入
model = face_model_mines.FaceModel(args)
 
##图片载入
path_img1 = '/data/tensorflow/WorkSpace/syd/FaceRecognition/Projects/insightface/deploy/test/shenyanding1.jpg'
path_img2 = '/data/tensorflow/WorkSpace/syd/FaceRecognition/Projects/insightface/deploy/test/shenyanding2.jpg'
 
#提取图片名称：姓名
s1 = path_img1.rfind('/', 1)+1
s2 = path_img1.rfind('.',1)-1
name_img1 = path_img1[s1:s2]
print("被检测人姓名：",name_img1)
 
ss1 = path_img2.rfind('/', 1)+1
ss2 = path_img2.rfind('.',1)-1
name_img2 = path_img2[ss1:ss2]
print("比对人姓名：",name_img2)
 
##根据模型计算人脸特征向量
img1 = cv2.imread(path_img1)
img1  = model.get_input(img1)
f1 = model.get_feature(img1)
img2 = cv2.imread(path_img2)
img2  = model.get_input(img2)
f2 = model.get_feature(img2 )
 
##计算余弦距离
dist1 = np.sum(np.square(f1-f2))
dist1 = float("%.2f" % dist1)
 
##输出模型检测结果
#检测逻辑：如果两张人脸的距离超过设定阈值则不是同一人，如果小于等于设定阈值则为同一人
dist_yuzhi = 1.24
 
if dist1<=dist_yuzhi:
    print("模型检测结果：是同一人")
else:
    print("模型检测结果：不是同一人")