Python使用Face++的人脸识别detect API进行本地图片情绪识别并存入excel

Posted 2020-11-02 ivywong

准备工作

首先，需要在Face++的主页注册一个账号，在控制台去获取API Key和API Secret。

然后在本地文件夹准备好要进行情绪识别的图片/相片。

 

代码

介绍下所使用的第三方库

——urllib2是使用各种协议完成打开url的一个库

——time是对时间进行处理的一个库，以下代码中其实就使用了sleep()和localtime()两个函数，sleep()是用来让程序暂停几秒的，localtime()是格式化时间戳为本地的时间

——xlwt是对excel进行写入操作的一个库

——os是操作系统的相关功能的一个库，例如用来处理文件和目录之类的

——json （Emmmmmm……我也不知道该怎么解释这个）

——PIL是Python图像处理库

  1 # -*- coding: utf-8 -*-
  2 # version:python2.7.13
  3 # author:Ivy Wong
  4 
  5 # 导入相关模块
  6 import urllib2
  7 import time, xlwt, os,json
  8 from PIL import Image
  9 
 10 # 使用face++的api识别情绪
 11 def useapi(img):
 12     http_url = ‘https://api-cn.faceplusplus.com/facepp/v3/detect‘
 13     boundary = ‘----------%s‘ % hex(int(time.time() * 1000))
 14     data = []
 15     data.append(‘--%s‘ % boundary)
 16     data.append(‘Content-Disposition: form-data; name="%s"\\r\\n‘ % ‘api_key‘)
 17     data.append(key)
 18     data.append(‘--%s‘ % boundary)
 19     data.append(‘Content-Disposition: form-data; name="%s"\\r\\n‘ % ‘api_secret‘)
 20     data.append(secret)
 21     data.append(‘--%s‘ % boundary)
 22     fr = open(img, ‘rb‘)
 23     data.append(‘Content-Disposition: form-data; name="%s"; filename=" "‘ % ‘image_file‘)
 24     data.append(‘Content-Type: %s\\r\\n‘ % ‘application/octet-stream‘)
 25     data.append(fr.read())
 26     fr.close()
 27     data.append(‘1‘)
 28     data.append(‘--%s‘ % boundary)
 29     data.append(‘Content-Disposition: form-data; name="%s"\\r\\n‘ % ‘return_attributes‘)
 30     data.append("gender,age,emotion,ethnicity")# 这里可以还可以返回其他参数，具体可以参看face++的api文档
 31     data.append(‘--%s--\\r\\n‘ % boundary)
 32 
 33     http_body = ‘\\r\\n‘.join(data)
 34     # build http request
 35     req = urllib2.Request(http_url)
 36     # header
 37     req.add_header(‘Content-Type‘, ‘multipart/form-data; boundary=%s‘ % boundary)
 38     req.add_data(http_body)
 39     try:
 40         resp = urllib2.urlopen(req,timeout=5)
 41         qrcont = json.load(resp)
 42     except urllib2.HTTPError as e:
 43         print e.read()
 44     return qrcont
 45 
 46 
 47 # 将json字典写入excel
 48 # 变量用来循环时控制写入单元格，感觉这种方式有点傻，但暂时想不到优化方法
 49 def writeexcel(img, worksheet, row, files_name):
 50     parsed = useapi(img)
 51     if not parsed[‘faces‘]:
 52         print ‘This picture do not have any face‘
 53     else:
 54         if len(parsed[‘faces‘])<=5: # 由于免费API限制，只能返回5张人脸信息
 55             for list_item in parsed[‘faces‘]:
 56                 # 写入文件名
 57                 filename, extension = os.path.splitext(files_name)
 58                 worksheet.write(row, 0, filename)
 59 
 60                 # 写入时间戳
 61                 daystamp, timestamp, hourstamp = gettimestamp(img)
 62                 worksheet.write(row, 1, label=daystamp)
 63                 worksheet.write(row, 2, label=timestamp)
 64                 worksheet.write(row, 3, hourstamp)
 65 
 66                 # 写入api返回的数据
 67                 emotion = []
 68                 for key1, value1 in list_item.items():
 69                     if key1 == ‘attributes‘:
 70                         for key2, value2 in value1.items():
 71                             if key2 == ‘age‘:
 72                                 worksheet.write(row, 5, value2[‘value‘])
 73                             elif key2 == ‘emotion‘:
 74                                 for key3, value3 in value2.items():
 75                                     if key3 == ‘sadness‘:
 76                                         worksheet.write(row, 8, value3)
 77                                         emotion.append(value3)
 78                                     elif key3 == ‘neutral‘:
 79                                         worksheet.write(row, 9, value3)
 80                                         emotion.append(value3)
 81                                     elif key3 == ‘disgust‘:
 82                                         worksheet.write(row, 10, value3)
 83                                         emotion.append(value3)
 84                                     elif key3 == ‘anger‘:
 85                                         worksheet.write(row, 11, value3)
 86                                         emotion.append(value3)
 87                                     elif key3 == ‘surprise‘:
 88                                         worksheet.write(row, 12, value3)
 89                                         emotion.append(value3)
 90                                     elif key3 == ‘fear‘:
 91                                         worksheet.write(row, 13, value3)
 92                                         emotion.append(value3)
 93                                     else:
 94                                         worksheet.write(row, 14, value3)
 95                                         emotion.append(value3)
 96                             elif key2 == ‘gender‘:
 97                                 worksheet.write(row, 6, value2[‘value‘])
 98                             elif key2 == ‘ethnicity‘:
 99                                 worksheet.write(row, 7, value2[‘value‘])
100                             else:
101                                 pass
102                     elif key1 == ‘face_token‘:
103                         worksheet.write(row, 4, value1)
104                     else:
105                         pass
106                 worksheet.write(row, 15, emotion.index(max(emotion)))
107 
108                 # 写入概率最大的情绪，0-neutral,1-sadness,2-disgust，3-anger，4-surprise，5-fear，6-happiness
109                 row += 1
110         else:
111             for list_item in parsed[‘faces‘][0:5]:
112                 # 写入文件名
113                 filename, extension = os.path.splitext(files_name)
114                 worksheet.write(row, 0, filename)
115 
116                 # 写入时间戳
117                 daystamp, timestamp, hourstamp = gettimestamp(img)
118                 worksheet.write(row, 1, label=daystamp)
119                 worksheet.write(row, 2, label=timestamp)
120                 worksheet.write(row, 3, hourstamp)
121 
122                 # 写入api返回的数据
123                 emotion = []
124                 for key1, value1 in list_item.items():
125                     if key1 == ‘attributes‘:
126                         for key2, value2 in value1.items():
127                             if key2 == ‘age‘:
128                                 worksheet.write(row, 5, value2[‘value‘])
129                             elif key2 == ‘emotion‘:
130                                 for key3, value3 in value2.items():
131                                     if key3 == ‘sadness‘:
132                                         worksheet.write(row, 8, value3)
133                                         emotion.append(value3)
134                                         print ‘1‘
135                                     elif key3 == ‘neutral‘:
136                                         worksheet.write(row, 9, value3)
137                                         emotion.append(value3)
138                                         print ‘2‘
139                                     elif key3 == ‘disgust‘:
140                                         worksheet.write(row, 10, value3)
141                                         emotion.append(value3)
142                                         print ‘3‘
143                                     elif key3 == ‘anger‘:
144                                         worksheet.write(row, 11, value3)
145                                         emotion.append(value3)
146                                         print ‘4‘
147                                     elif key3 == ‘surprise‘:
148                                         worksheet.write(row, 12, value3)
149                                         emotion.append(value3)
150                                         print ‘5‘
151                                     elif key3 == ‘fear‘:
152                                         worksheet.write(row, 13, value3)
153                                         emotion.append(value3)
154                                         print ‘6‘
155                                     else:
156                                         worksheet.write(row, 14, value3)
157                                         emotion.append(value3)
158                                         print ‘7‘
159                             elif key2 == ‘gender‘:
160                                 worksheet.write(row, 6, value2[‘value‘])
161                             elif key2 == ‘ethnicity‘:
162                                 worksheet.write(row, 7, value2[‘value‘])
163                             else:
164                                 pass
165                     elif key1 == ‘face_token‘:
166                         worksheet.write(row, 4, value1)
167                     else:
168                         pass
169                 worksheet.write(row, 15, emotion.index(max(emotion)))
170                 # 写入概率最大的情绪，0-neutral,1-sadness,2-disgust，3-anger，4-surprise，5-fear，6-happiness
171                 row += 1
172                 print ‘Success! The pic ‘ + str(files_name) + ‘ was detected!‘
173 
174     return row, worksheet
175 
176 # 获取图片大小
177 def imagesize(img):
178     Img = Image.open(img)
179     w, h = Img.size
180     return w,h
181 
182 # 获取时间戳
183 def gettimestamp(path):
184     statinfo = os.stat(path)
185     timeinfo = time.localtime(statinfo.st_ctime)
186     daystamp = str(timeinfo.tm_year) + ‘-‘ + str(timeinfo.tm_mon) + ‘-‘ + str(timeinfo.tm_mday)
187     timestamp = str(timeinfo.tm_hour) + ‘:‘ + str(timeinfo.tm_min) + ‘:‘ + str(timeinfo.tm_sec)
188     hourstamp = timeinfo.tm_hour + timeinfo.tm_min / 60.0 + timeinfo.tm_sec / 3600.0
189     return daystamp, timestamp, hourstamp
190 
191 
192 key = "your_key"
193 secret = "your_secret"
194 path = r"图片文件夹路径"
195 # 注意：由于我是对同一文件夹下的多个文件夹中的图片进行识别，所以这个path是图片所在文件夹的上一级文件夹。文件夹名尽量使用英文与数字，不然可能因为编码问题报错
196 
197 # 创建excel
198 workbook = xlwt.Workbook(encoding=‘utf-8‘)
199 
200 for root, dirs, files in os.walk(path, topdown=False):
201     for folder in dirs:
202         print ‘Let us start dealing with folder ‘ + folder
203 
204         # 创建一个新的sheet
205         worksheet = workbook.add_sheet(folder)
206         # 设置表头
207         title = [‘PhotoID‘, ‘daystamp‘, ‘timestamp‘, ‘hourstamp‘,‘faceID‘, ‘age‘, ‘gender‘, ‘ethnicity‘, ‘sadness‘,
208                  ‘neutral‘,‘disgust‘, ‘anger‘, ‘surprise‘, ‘fear‘, ‘happiness‘, ‘emotion‘]
209         for col in range(len(title)):
210             worksheet.write(0, col, title[col])
211 
212         # 遍历每个folder里的图片
213         row = 1
214         for root2, dirs2, files2 in os.walk(path + ‘\\\\‘ + folder):
215             for files_name in files2:
216                 img = path + ‘\\\\‘ + folder + ‘\\\\‘ + files_name
217                 try:
218                     print ‘Now, the program is going to deal with ‘ + folder + ‘ pic‘ + str(files_name)
219                     w,h=imagesize(img)
220                     if w<48 or h<48 or w>4096 or h>4096:#API对图片大小的限制
221                         print ‘invalid image size‘
222                     else:
223                         row, worksheet = writeexcel(img, worksheet, row, files_name)
224 
225                 except:
226                     print ‘超过了并发数！等一下！‘
227                     time.sleep(3)
228                     print ‘The program is going to work‘
229                     print ‘Now, the program is going to deal with ‘ + folder + ‘ pic‘ + str(files_name)
230                     row, worksheet = writeexcel(img, worksheet, row, files_name)
231 
232 
233         workbook.save(‘detactface_facepp_flickr.xls‘)
234         print ‘The current folder is done.‘
235 
236 # 保存文件
237 workbook.save(‘detectface.xls‘)
238 print ‘All done!‘

 

成果

最后生成的excel大概是这个样子。

其中emotion就是概率最大的情绪，0-neutral,1-sadness,2-disgust，3-anger，4-surprise，5-fear，6-happiness。

 

探讨

在我自己运行过程中发现了一个问题，由于使用的是免费API，有并发限制，多次超过并发数，urlopen就会返回403。于是try失败，运行except，return时无定义的qrcont而报错。

1     try:
2         resp = urllib2.urlopen(req,timeout=5)
3         qrcont = json.load(resp)
4     except urllib2.HTTPError as e:
5         print e.read()
6     return qrcont

这就非常尴尬了，所以目前基本上都是大晚上在用这个代码跑……不知看到的各位有何高见？