This project is for automatically collecting and analysing photos within a wechat chatgroup.
- wxpy : serves as a robot to collect photos and reply messages
- mongoengine : provides interaction with the database
- (private) C++ server: provides face detection and recognition
- Baidu Face API: provides face detection and attributes analysis
We provide development kit to parse text messages as commands that request some statistic data. To develop your own parser, you could follow these steps:
- Create a file in
processors/folder (e.g.LikeTextProcessor.py) - Declare a class as the subclass of TextProcessor:
from TextParser import TextParser
class LikeTextProcessor(TextParser):
def __init__(self):
# register a reg-expression here for the command
super(LikeTextProcessor, self).__init__(r'@(.*)\s最中意谁')
def process(self, match):
# match is the match for the reg-expression
# DO your statistics
self.msg.reply('your reply text')
self.msg.reply_img('your image path') # you can get some image from the database and edit it with cv2- Register the class in the module init file
processors/__init__.pyby first import it and then add the name to__all__array.
from .LikeTextProcessor import LikeTextProcessor
__all__= [
'LikeTextProcessor',
]4 Test your code by running
$ python3 main.pyand input your test command.
The database format is as declared in models.py.
One document in this collection corresponds to one uploaded image.
date field is the upload time.
user field is the photographer (wechat name).
md5 field is used to exclude duplicate images.
img field is the raw image in binary.
stats field is the raw json of the analysis result, such as:
{"result": [
{"gender": "male", "pitch": 1.0520080327988, "race": "yellow", "location": {"top": 328, "left": 183, "height": 68, "width": 79}, "expression_probablity": 0.8709254860878, "race_probability": 0.99973601102829, "rotation_angle": -2, "gender_probability": 0.99991512298584, "glasses": 1, "age": 32.630474090576, "roll": -4.9296202659607, "identity": "yishan.zhang", "glasses_probability": 0.99987423419952, "face_probability": 1, "beauty": 26.279113769531, "yaw": -13.995768547058, "expression": 1, "faceshape": [{"probability": 0.053247287869453, "type": "square"}, {"probability": 0.05485212802887, "type": "triangle"}, {"probability": 0.10603010654449, "type": "oval"}, {"probability": 0.0053467354737222, "type": "heart"}, {"probability": 0.78052371740341, "type": "round"}]},
{"gender": "female", "pitch": 0.82871550321579, "race": "yellow", "location": {"top": 485, "left": 15, "height": 94, "width": 101}, "expression_probablity": 0.98333412408829, "race_probability": 0.99998557567596, "rotation_angle": 3, "gender_probability": 0.99999010562897, "glasses": 1, "age": 25.832576751709, "roll": 1.7325274944305, "identity": "yuting.liu", "glasses_probability": 0.9995750784874, "face_probability": 1, "beauty": 44.97737121582, "yaw": -13.487314224243, "expression": 1, "faceshape": [{"probability": 0.048728778958321, "type": "square"}, {"probability": 0.20114889740944, "type": "triangle"}, {"probability": 0.017348473891616, "type": "oval"}, {"probability": 0.039832547307014, "type": "heart"}, {"probability": 0.69294130802155, "type": "round"}]},
{"identity": "ding.liu", "location": {"top": 179, "left": 458, "height": 41, "width": 41}},
{"gender": "male", "pitch": 2.9942271709442, "race": "yellow", "location": {"top": 442, "left": 1170, "height": 110, "width": 121}, "expression_probablity": 0.99987399578094, "race_probability": 0.99999988079071, "rotation_angle": 6, "gender_probability": 0.99999988079071, "glasses": 0, "age": 29.096225738525, "roll": 7.8307189941406, "identity": "shaoyan.sun", "glasses_probability": 0.99998927116394, "face_probability": 1, "beauty": 40.45418548584, "yaw": 8.522575378418, "expression": 1, "faceshape": [{"probability": 0.021569181233644, "type": "square"}, {"probability": 0.59955924749374, "type": "triangle"}, {"probability": 0.046120498329401, "type": "oval"}, {"probability": 0.010387846268713, "type": "heart"}, {"probability": 0.3223631978035, "type": "round"}]},
{"gender": "male", "pitch": -4.6260681152344, "race": "yellow", "location": {"top": 321, "left": 912, "height": 71, "width": 79}, "expression_probablity": 0.99428832530975, "race_probability": 0.9999783039093, "rotation_angle": -10, "gender_probability": 0.99998164176941, "glasses": 1, "age": 27.871730804443, "roll": -7.0358691215515, "identity": "tingcheng.wu", "glasses_probability": 0.99929440021515, "face_probability": 1, "beauty": 40.989395141602, "yaw": 18.345357894897, "expression": 0, "faceshape": [{"probability": 0.28823563456535, "type": "square"}, {"probability": 0.0194001365453, "type": "triangle"}, {"probability": 0.2190715521574, "type": "oval"}, {"probability": 0.0068779760040343, "type": "heart"}, {"probability": 0.4664146900177, "type": "round"}]}
],
"result_num": 5}stats may not exist because no analysis results are provided.
The result in stats.result may not contain attributes, and the identity may be empty string because the face is not recognized.
This is mainly for querying the user name according the the identity.
identity = "kai.yu"
try:
user = Users.objects.get(identity=identity)
print(user.name)
except Users.DoesNotExist:
passShame that this collection name is a typo...
The occurrences of people.
identity can be "unknown".
Attributes related fields may be empty, but location and img always exists.
PLEASE test with main.py rather than bot.py which is for production use, because frequent request of wechat login may cause account locking!!!
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent