Aliyun_人脸人体_人脸检测与五官定位
import oss2
import json
import cv2
import matplotlib.pyplot as plt
%matplotlib inline
import numpy as np
AccessKeyId="xxxxx"
AccessKeySecret="xxxx"
auth = oss2.Auth(AccessKeyId, AccessKeySecret)
bucket = oss2.Bucket(auth, 'https://oss-cn-shanghai.aliyuncs.com', 'ocr-yjs')
from aliyunsdkcore.client import AcsClient
from aliyunsdkcore.acs_exception.exceptions import ClientException
from aliyunsdkcore.acs_exception.exceptions import ServerException
from aliyunsdkcore.auth.credentials import AccessKeyCredential
from aliyunsdkcore.auth.credentials import StsTokenCredential
credentials = AccessKeyCredential(AccessKeyId, AccessKeySecret)
client = AcsClient(region_id='cn-shanghai', credential=credentials)
import seaborn as sns
"""
https://help.aliyun.com/document_detail/151969.html
人脸检测与五官定位
"""
from aliyunsdkfacebody.request.v20191230.DetectFaceRequest import DetectFaceRequest
request = DetectFaceRequest()
request.set_accept_format('json')
filepath="images/test2.jpg"
with open(filepath, 'rb') as fileobj:
bucket.put_object(filepath, fileobj)
url="https://ocr-yjs.oss-cn-shanghai.aliyuncs.com/"+filepath
request.set_ImageURL(url)
response = client.do_action_with_exception(request)
response=json.loads(str(response, encoding='utf-8'))
print(response)
{'RequestId': '9C82A124-9FFD-5723-947C-91043D584A1A', 'Data': {'FaceProbabilityList': [0.9713473320007324], 'LandmarkCount': 105, 'FaceRectangles': [75, 88, 133, 173], 'Landmarks': [73.33972930908203,
248.45582580566406],
'Qualities': {'ScoreList': [90.26331329345703],
'NoiseList': [99.47721099853516],
'BlurList': [99.99376678466797],
'FnfList': [100.0],
'GlassList': [99.98408508300781],
'MaskList': [86.59868621826172],
'IlluList': [99.99360656738281],
'PoseList': [7.566176891326904]},
'FaceCount': 1,
'Pupils': [94.2857437133789,
161.79794311523438,
4.954795837402344,
152.37928771972656,
155.88494873046875,
4.954795837402344],
'PoseList': [-17.756900787353516, -6.129109859466553, -6.328291893005371]}
FaceProbabilityList=Data["FaceProbabilityList"]
FaceProbabilityList # 截图中包含人脸的概率，取值范围0~1。如有多个人脸，则依次返回
[0.9713473320007324]
FaceRectangles=Data["FaceRectangles"]
FaceRectangles # 返回人脸矩形框，分别是[left, top, width, height]。如有多个人脸，则依次顺延，返回矩形框。例如有两个人脸则返回
# [left1, top1, width1, height1, left2, top2, width2, height2]
left=FaceRectangles[0]
top=FaceRectangles[1]
width=FaceRectangles[2]
height=FaceRectangles[3]
img=plt.imread(filepath)
res=cv2.rectangle(img,(left,top), (left+width,top+height),(0,0,255), 4)
plt.imshow(res)
LandmarkCount=Data["LandmarkCount"]
LandmarkCount # 人脸特征点数目，目前固定为105点。依次为：眉毛24点，眼睛32点，鼻子6点，嘴巴34点，外轮廓9点。
105
Landmarks=Data['Landmarks']
Landmarks
[73.33972930908203,
....................................
209.629075,
248.45582580566406]
Landmarks
# 人脸特征点数目，目前固定为105点。依次为：眉毛24点，眼睛32点，鼻子6点，嘴巴34点，外轮廓9点。
[73.33972930908203,
145.13458251953125,
.....................
209.62904357910156,
185.9634552001953,
108.64936828613281,
242.60096740722656,
169.647216796875,
248.45582580566406]
def get_X_Y(data):
list2=[]
for index,value in enumerate(data):
if index%2==0:
list2.append([int(data[index]),int(data[index+1])])
return list2
Eyebrows=get_X_Y(Landmarks[:48])
Eyes=get_X_Y(Landmarks[48:112])
Nose=get_X_Y(Landmarks[112:124])
Mouth=get_X_Y(Landmarks[124:192])
Outercontour=get_X_Y(Landmarks[192:])
colors=[[ int(i) for i in [ i*255 for i in color]] for color in sns.color_palette("hls",5)]
colors
[[219, 94, 86], [184, 219, 86], [86, 219, 147], [86, 131, 219], [200, 86, 219]]
img=plt.imread(filepath)
for lines in Eyebrows:
# {'X': 113, 'Y': 225}
center=(lines[0],lines[1])
cv2.circle(img, center, 1, colors[0])
plt.imshow(img)
for lines in Eyes:
# {'X': 113, 'Y': 225}
center=(lines[0],lines[1])
cv2.circle(img, center, 1, colors[1])
plt.imshow(img)
for lines in Nose:
# {'X': 113, 'Y': 225}
center=(lines[0],lines[1])
cv2.circle(img, center, 1, (0,0,0))
plt.imshow(img)
for lines in Mouth:
# {'X': 113, 'Y': 225}
center=(lines[0],lines[1])
cv2.circle(img, center, 1, (0,0,0))
plt.imshow(img)
for lines in Outercontour:
# {'X': 113, 'Y': 225}
center=(lines[0],lines[1])
cv2.circle(img, center, 1, (255,255,255))
plt.imshow(img)
"""
'Qualities': 人脸质量情况，分数越高表示越有利于识别。
{'ScoreList': [90.26331329345703],
'NoiseList': [99.47721099853516],
'BlurList': [99.99376678466797],
'FnfList': [100.0],
'GlassList': [99.98408508300781],
'MaskList': [86.59868621826172],
'IlluList': [99.99360656738281],
'PoseList': [7.566176891326904]
},
'FaceCount': 1, 检测出的人脸个数。
'Pupils': 左右两个瞳孔的中心点坐标和半径
[94.2857437133789,
161.79794311523438,
4.954795837402344,
152.37928771972656,
155.88494873046875,
4.954795837402344],
'PoseList': [-17.756900787353516, -6.129109859466553, -6.328291893005371]
返回人脸姿态，格式为[yaw, pitch, roll]。如有多个人脸，则依次顺延。
yaw为左右角度，取值范围-90~90。
pitch为上下角度，取值范围-90~90。
roll为平面旋转角度，取值范围-180~180。
"""