def update_uncentain_data(meta_info, batch_id=0):
"""
[
summary
]
Args:
meta_info ([type]): [description]
batch_id (int, optional): [description]. Defaults to 0.
"""
data_oid = meta_info["data_oid"]
# todo 米桂田.
for db in database.get_db():
query_result_list = db.query(database.XUncertainCategoryTable).filter(
database.XUncertainCategoryTable.data_oid == data_oid).filter(
database.XUncertainCategoryTable.batch_id == batch_id)
if query_result_list.count() > 0:
rule_predict_ret_obj = query_result_list.one()
rule_predict_ret_obj.category = meta_info["category"]
rule_predict_ret_obj.category_from = meta_info["category_from"]
rule_predict_ret_obj.prob = meta_info["prob"]
rule_predict_ret_obj.finished = 0
rule_predict_ret_obj.judgment_type = 0
rule_predict_ret_obj.judgment_category = "null"
else:
rule_predict_ret_obj = database.XUncertainCategoryTable(
batch_id=batch_id,
data_oid=data_oid,
category=meta_info["category"],
prob=meta_info["prob"],
category_from=meta_info["category_from"])
db.add(rule_predict_ret_obj)
db.commit()
pass
def unknown_faces():
db = get_db()
images = faces.find_unknown_in_db(cursor=db.cursor, api=api)
existing_people = get_person().list()
return render_template('unknown_faces.html.jinja',
images=images,
people=existing_people,
crop_size=100)
def get_ai_prediction_result(batch_id=0, ai_id=0):
"""[summary]
Args:
batch_id (int, optional): [批量号码]. Defaults to 0.
ai_id (int, optional): [模型预测密码]. Defaults to 0.
Returns:
List[database.AiPredictonOfResultsTable]: [
[{
"data_id":1,
"category":"",
"proba":0.2,
}]
]
"""
ans_data = {}
# ans_data={
# "data_oid":{},
# }
for db in database.get_db():
data_list = db.query(
database.AiPredictonOfResultsTable.data_oid,
database.AiPredictonOfResultsTable.meta_info,
database.AiPredictonOfResultsTable.category).filter(
database.AiPredictonOfResultsTable.batch_id == batch_id
).filter(database.AiPredictonOfResultsTable.ai_id == ai_id)
metric_info = get_metric_info(ai_id, "ai_{}".format(ai_id))
for data in data_list:
data_info_dict = json.loads(data.meta_info)
proba = 0.1
try:
data_info_dict = json.loads(data.meta_info)
category_id = data_info_dict["category_id"]
# 这一步用来计算可能,数据可能是这一类真正的概率.
#
category = data_info_dict["category"]
current_proba = data_info_dict["proba_list"][category_id -
1] # 获取概率
precision = metric_info["precision"].get(category, 1)
proba = current_proba * precision
except Exception as identifier:
print(identifier)
ans_data[data.data_oid] = {
"data_oid": data.data_oid,
"category": data.category,
"category_id": category_id,
"proba": proba
}
return ans_data
def start_scan():
db = get_db()
if threads['photos'] == None or not threads['photos'].is_alive():
queue = PhotoQueue(db.cnx)
queue.fill_queue()
threads['photos'] = Thread(target=queue.consume,
kwargs=dict(process=process_photos.process(
api=api, cursor=db.cnx.cursor())))
threads['photos'].start()
return {'result': 'Scan has started'}
return {'result': 'in_progress'}
def addUser():
conn = get_db()
c = conn.cursor()
print("Opened database successfully")
# 业务逻辑
c.execute("INSERT INTO user (ID,USERNAME,password) \
VALUES (1, 'Paul', '12321' )")
conn.commit()
print("Records created successfully")
conn.close()
return 'add success'
def updateUser():
conn = get_db()
c = conn.cursor()
print("Opened database successfully")
# 业务逻辑
c.execute("UPDATE user set username = '******' where ID=1")
conn.commit()
print("Total number of rows updated :", conn.total_changes)
print("Records created successfully")
conn.close()
return 'update success'
def getUser():
conn = get_db()
c = conn.cursor()
print("Opened database successfully")
# 业务逻辑
cursor = c.execute("SELECT id, username, password from user")
for row in cursor:
print(row[1])
print(row["password"])
print("Records created successfully")
conn.close()
return 'get success'
def possible_faces_matches():
db = get_db()
ids = [int(id) for id in request.args.getlist('id')]
print('Looking for', ids)
faces_encodings = faces.get_faces_encodings(faces_ids=ids,
cursor=db.cursor)
possible_faces = faces.find_closest_matches_in_db(
faces_encodings=faces_encodings,
ignore_known=True,
exclude_list=ids,
cursor=db.cnx.cursor(dictionary=True))
result = list(map(format_faces_response, possible_faces))
return {'result': {'faces': result}}
def possible_face_matches(id):
db = get_db()
face_encodings = faces.get_face_encodings(face_id=id, cursor=db.cursor)
possible_faces = faces.find_closest_match_in_db(
face_encodings=face_encodings,
ignore_known=True,
exclude_list=(id, ),
cursor=db.cnx.cursor(dictionary=True))
return {
'result': {
'id': id,
'faces': list(map(format_faces_response, possible_faces))
}
} @ app.route('/faces/<int:id>/matches')
def get_rule_model_group(batch_id: str) -> dict:
"""
根据batch_id,获得此批次的规则模型组，供加载使用
Args:
batch_id (int): 批次ID
Returns:
{category：{rule_model}}
"""
if not isinstance(batch_id, str):
batch_id = str(batch_id)
rule_model_group = {}
category_list = []
for db in database.get_db():
query_categry_list = db.query(
database.BatchRegularInfoTable.category_id).filter(
database.BatchRegularInfoTable.batch_id == batch_id).group_by(
database.BatchRegularInfoTable.category_id)
if query_categry_list.count() == 0:
return rule_model_group #TODO: 返回是空如何处理
else:
for query_categry in query_categry_list:
category_list.append(query_categry[0])
for category_id in category_list:
pos_patterns_list = []
neg_patterns_list = []
query_result_list = db.query(
database.BatchRegularInfoTable).filter(
database.BatchRegularInfoTable.batch_id == batch_id,
database.BatchRegularInfoTable.category_id == category_id)
for query_result in query_result_list:
if query_result.regular_type == 'pos':
pos_patterns_list.append(query_result.content)
elif query_result.regular_type == 'neg':
neg_patterns_list.append(query_result.content)
else:
pass #TODO: 是否还会有其他类型的规则，是否需要处理
category = db.query(
database.BatchCategoryInfoTable.category).filter(
database.BatchCategoryInfoTable.category_id ==
category_id).one()
rule_model_group[category[0]] = {
"origin_pos_patterns": pos_patterns_list,
"origin_neg_patterns": neg_patterns_list
}
return rule_model_group
def update_predict_data_rule(meta_info, batch_id=0, model_id=0):
# 更新一下数据
"""
[
]
Args:
meta_info ([dict]): meta_info ([dict]):
{
"category_id": 2,
"category": "预计的业绩",
"data_oid": 7
}
batch_id (int, optional): [description]. Defaults to 0.
ai_id (int, optional): [description]. Defaults to 0.
"""
data_oid = meta_info["data_oid"]
for db in database.get_db():
query_result_list = db.query(
database.RulePredictionOfResultsTable).filter(
database.RulePredictionOfResultsTable.batch_id ==
batch_id).filter(
database.RulePredictionOfResultsTable.model_id == model_id
).filter(
database.RulePredictionOfResultsTable.data_oid == data_oid)
str_meta_info = json.dumps(meta_info,
ensure_ascii=False,
indent=4,
cls=NpEncoder)
if query_result_list.count() > 0:
rule_predict_ret_obj = query_result_list.one()
db_meta_info = json.loads(rule_predict_ret_obj.meta_info)
print(db_meta_info)
if meta_info["category"] != rule_predict_ret_obj.category:
rule_predict_ret_obj.category = meta_info["category"]
rule_predict_ret_obj.meta_info = str_meta_info
db.commit() # update data in database
else:
rule_predict_ret_obj = database.RulePredictionOfResultsTable(
data_oid=data_oid,
category=meta_info["category"],
batch_id=batch_id,
model_id=model_id,
meta_info=str_meta_info)
db.add(rule_predict_ret_obj)
db.commit()
def update_predict_data(meta_info, batch_id=0, ai_id=0):
"""[summary]
Args:
data_id ([int]): [description]
meta_info ([dict]):
{
"category_id": 2,
"decision_list": [
0.849419160098665,
3.2599307808093747,
2.118750077455952,
-0.25528040257805545
],
"proba_list": [
0.0013409690326336684,
0.9808453085781627,
0.01730583467187295,
0.0005078877173305729
],
"category": "预计的业绩",
"data_oid": 7
}
batch_id (int, optional): [description]. Defaults to 0.
ai_id (int, optional): [description]. Defaults to 0.
Returns:
[type]: [description]
"""
data_oid = meta_info["data_oid"]
for db in database.get_db():
query_result_list = db.query(
database.AiPredictonOfResultsTable).filter(
database.AiPredictonOfResultsTable.batch_id == batch_id
).filter(database.AiPredictonOfResultsTable.ai_id == ai_id).filter(
database.AiPredictonOfResultsTable.data_oid == data_oid)
str_meta_info = json.dumps(meta_info,
ensure_ascii=False,
indent=4,
cls=NpEncoder)
if query_result_list.count() > 0:
ai_predict_ret_obj = query_result_list.one()
ai_predict_ret_obj.meta_info
db_meta_info = json.loads(ai_predict_ret_obj.meta_info)
#print(db_meta_info)
if meta_info["category"] != db_meta_info["category"]:
ai_predict_ret_obj.category = meta_info["category"]
ai_predict_ret_obj.meta_info = str_meta_info
db.commit() # update data in database
else:
ai_ret_obj = database.AiPredictonOfResultsTable(
data_oid=data_oid,
category=meta_info["category"],
batch_id=batch_id,
ai_id=ai_id,
meta_info=str_meta_info)
db.add(ai_ret_obj)
db.commit()
return 0
return 1
#print(list_vec)
pass
def get_person():
db = get_db()
person = People(cnx=db.cnx)
return person
def unasign_face(id):
db = get_db()
faces.unlink_from_person(id=id, cursor=db.cursor, cnx=db.cnx)
return {'result': {'id': id, 'ok': True}}
