def put(self):
try:
name = get_argument('username', required=True, help='请输入用户名')
email = get_argument('email', required=False, help='请输入邮箱')
user1 = users.User.query.filter_by(name).first()
user1.name = email
db.session.commit()
return make_response(status=HttpStatus.SUCCESS)
except:
db.session.rollback()
raise AiException()
def get(self):
name = get_argument('username', required=True, help='请输入用户名')
password = get_argument('password', required=True, help='请输入密码')
# Get请求也可通过如下方式获取参数
# name = request.args.get('username', '')
# password = request.args.get('password', '')
logger.info("name:%s, password:%s" % (name, password))
return make_response(status=HttpStatus.SUCCESS,
data={
"name": name,
"password": password
})
def post(self):
name = get_argument('username', required=True, help='请输入用户名')
password = get_argument('password', required=False, help='请输入密码')
#也可直接通过rest.form来获取请求报文
# name = request.form['username']
# password = request.form['password']
logger.info("name:%s, password:%s" % (name, password))
return make_response(status=HttpStatus.SUCCESS,
data={
"name": name,
"password": password
})
def post(self):
path = get_argument(constants.cons_request_path,
required=True,
help='文件路径为空')
if os.path.isfile(path):
img = cv2.imread(path)
# cv2.imwrite("src.jpg", img)
# 将图象由OpenCv的BGR转成RGB格式，wuyunzhen.zh, 20190309
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# cv2.imwrite("rgb.jpg", img)
tmp_embs = humanRecognizeSrv.get_embedding(img)
if len(tmp_embs) > 0:
logger.info("计算人脸特征:")
emb = tmp_embs[0].reshape(-1)
# 将ndArray转byte，再将byte转字符串
emb_list = emb.tolist()
data = ",".join(str(s) for s in emb_list)
ai_print(data)
return make_response(status=HttpStatus.SUCCESS,
data={"rslt_cmnt": data})
raise AiException(HttpStatus.SERVER_ERROR, AiErrorCode.YCEA4021009,
path)
else:
raise AiException(HttpStatus.SERVER_ERROR, AiErrorCode.YCEA4021002,
path)
def post(self):
ccbins_id = get_argument(constants.cons_request_ccbins_id,
required=True,
help='机构编号不能为空')
# 已改成后台自动定时读取人脸库信息
# humanRecognizeSrv.reload_faces_by_ins(ccbins_id)
return make_response(status=HttpStatus.SUCCESS)
def __check_args(self):
get_argument(constants.cons_request_appid,
required=True,
help='应用组件编号不能为空')
get_argument(constants.cons_request_trace_id,
required=True,
help='全局追踪号不能为空')
ccbins_id = get_argument(constants.cons_request_ccbins_id,
required=True,
help='机构编号不能为空')
empe_Grp = get_argument(constants.cons_request_empe_Grp,
type=list,
required=True,
help='人员信息不能为空')
empe_infs = []
for empe in empe_Grp:
empe_inf = {}
empe_inf[constants.cons_request_ccbins_id] = ccbins_id
empe_inf[constants.cons_request_empe_id] = empe[
constants.cons_request_empe_id]
empe_inf[constants.cons_request_usr_nm] = empe[
constants.cons_request_usr_nm]
empe_inf[constants.cons_request_mnt_tp_cd] = empe[
constants.cons_request_mnt_tp_cd]
if os.path.isfile(empe[constants.cons_request_path]):
empe_inf[constants.cons_request_img] = cv2.imread(
empe[constants.cons_request_path])
else:
raise AiException(HttpStatus.SERVER_ERROR,
AiErrorCode.YCEA4021002,
empe[constants.cons_request_path])
empe_infs.append(empe_inf)
return empe_infs
def get(self):
city = get_argument('city', required=True, help='请输入城市')
if city is None:
city = '武汉'
# 下面接口须连互联网，查天气的接口
url = "https://www.sojson.com/open/api/weather/json.shtml"
request_data = {"city": city}
headers = {
"User-Agent":
"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.36"
}
ret = requests.get(url, params=request_data, headers=headers)
# ### ### print(ret.url)
if ret.status_code == 200:
result = json.loads(ret.text)
return make_response(status=HttpStatus.SUCCESS, data=result)
return make_response(status=HttpStatus.SERVER_ERROR)
def post(self):
### print(request.__dict__)
image_base64 = get_argument('image_base64', required=True, help='图片信息')
# # 将base64转成文件
img_data = base64.b64decode(image_base64)
# 转换为np数组
img_array = np.fromstring(img_data, np.uint8)
# 转换成opencv可用格式
img_data = cv2.imdecode(img_array, cv2.IMREAD_COLOR)
# ### print(image_base64)
# img_data = np.asarray(image_base64)
dir = current_app.config.get('BASE_DIR')
path = dir + "/image/" + datetime.datetime.now().strftime(
'%Y%m%d_%H%M%S_%f') + ".jpg"
cv2.imwrite(path, img_data)
return make_response(status=HttpStatus.SUCCESS)
def post(self):
ccbinsId = get_argument("ccbinsId", required=True, help='所属机构不能为空')
indexCode = get_argument("indexCode", required=True, help='所属设备不能为空')
status = get_argument("status", required=True, help='报警类型不能为空')
cct_rcrd_inf = get_argument("cct_rcrd_inf", required=False)
eqmt_id = get_argument("eqmt_id", required=False)
use_width = get_argument("use_width", required=False)
thresh = get_argument("thresh", required=False)
time_step = get_argument("time_step", required=False)
video_conn_info_po = VideoConnInfo()
video_conn_info_po.id = gen_uuid()
video_conn_info_po.ccbins_id = ccbinsId
video_conn_info_po.cct_rcrd_inf = cct_rcrd_inf
video_conn_info_po.eqmt_id = eqmt_id
video_conn_info_po.index_code = indexCode
video_conn_info_po.status = status
video_conn_info_po.param1 = use_width
video_conn_info_po.param2 = thresh
video_conn_info_po.param3 = time_step
video_conn_dao.add_record(video_conn_info_po)
return make_response(status=HttpStatus.SUCCESS,
data={"code": "success"})
def main():
args = set_default(get_argument())
args, initial_epoch = search_same(args)
if initial_epoch == -1:
# training was already finished!
return
elif initial_epoch == 0:
# first training or training with snapshot
weekday = ["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"]
temp = datetime.now()
args.stamp = "{:02d}{:02d}{:02d}_{}_{:02d}_{:02d}_{:02d}".format(
temp.year // 100,
temp.month,
temp.day,
weekday[temp.weekday()],
temp.hour,
temp.minute,
temp.second,
)
get_session(args)
logger = get_logger("MyLogger")
for k, v in vars(args).items():
logger.info("{} : {}".format(k, v))
##########################
# Generator
##########################
trainset, valset = set_dataset(args, logger)
train_generator = create_generator(args, trainset, "train", args.batch_size)
# for t in train_generator:
# print(sorted(t[1]['main_output'].numpy().argmax(axis=0)))
# print(t[0]['main_input'].shape, t[0]['main_input'].numpy().min(), t[0]['main_input'].numpy().max(), t[1]['main_output'])
val_generator = create_generator(args, valset, "val", args.batch_size)
# for t in val_generator:
# print(t[0][0].shape, t[0][1], t[1])
test_generator1 = create_generator(args, trainset, "val", 1)
test_generator2 = create_generator(args, valset, "val", 1)
# for t in test_generator2:
# print(t[0]['main_input'].shape, t[0]['arcface_input'])
if args.class_weight:
assert args.classes > 1
from sklearn.utils.class_weight import compute_class_weight
train_label = trainset[:, 1:].astype(np.int).argmax(axis=1)
class_weight = compute_class_weight(
class_weight="balanced", classes=np.unique(train_label), y=train_label
)
else:
class_weight = None
logger.info("TOTAL STEPS OF DATASET FOR TRAINING")
logger.info("========== trainset ==========")
steps_per_epoch = args.steps or len(trainset) // args.batch_size
logger.info(" --> {}".format(steps_per_epoch))
# logger.info(" --> {}".format(trainset[:, 2:].sum(axis=0)))
# logger.info(" --> {}".format(class_weight))
logger.info("=========== valset ===========")
validation_steps = len(valset) // args.batch_size
logger.info(" --> {}".format(validation_steps))
# logger.info(" --> {}".format(valset[:, 2:].sum(axis=0)))
##########################
# Model
##########################
model = create_model(args, logger)
if args.summary:
model.summary()
print(model.inputs[0])
print(model.get_layer(name="fc2"))
return
model = compile_model(args, model, steps_per_epoch)
logger.info("Build model!")
##########################
# Callbacks
##########################
callbacks = create_callbacks(args, test_generator1, test_generator2, trainset, valset)
logger.info("Build callbacks!")
##########################
# Train
##########################
model.fit(
x=train_generator,
epochs=args.epochs,
callbacks=callbacks,
validation_data=val_generator,
steps_per_epoch=steps_per_epoch,
validation_steps=validation_steps,
class_weight=class_weight,
initial_epoch=initial_epoch,
verbose=args.verbose,
)
def __check_args(self):
get_argument(constants.cons_request_appid,
required=True,
help='应用组件编号不能为空')
get_argument(constants.cons_request_trace_id,
required=True,
help='全局追踪号不能为空')
get_argument(constants.cons_request_content_type,
required=True,
help='文件类型不能为空')
get_argument(constants.cons_request_function,
required=True,
help='功能不能为空')
get_argument(constants.cons_request_ccbins_id,
required=True,
help='机构编号不能为空')
model_type = get_argument(constants.cons_request_model_type,
type=list,
required=True,
help='模型服务类型不能为空')
if not model_type:
raise AiException(HttpStatus.SERVER_ERROR, AiErrorCode.YCEA4021008)
path_list = get_argument(constants.cons_request_paths, required=False)
img_list = get_argument(constants.cons_request_imgs, required=False)
cv_image_list = []
# 从路径数组中读图
path_flag = False
if path_list:
path_flag = True
logger.debug('path字段非空，从路径信息' + ':'.join(path_list))
for path in path_list:
if os.path.isfile(path):
img = cv2.imread(path)
cv_image_list.append(img)
else:
raise AiException(HttpStatus.SERVER_ERROR,
AiErrorCode.YCEA4021002, path)
# 如果路径数组为空，从base64数组中读图
elif img_list:
logger.debug('path字段为空，取base64图片')
for base64_img in img_list:
if base64_img:
img_data = base64.b64decode(base64_img)
# 转换为np数组
img_array = np.fromstring(img_data, np.uint8)
# 转换成opencv可用格式
img = cv2.imdecode(img_array, cv2.IMREAD_COLOR)
cv_image_list.append(img)
else:
raise AiException(HttpStatus.SERVER_ERROR,
AiErrorCode.YCEA4021006)
else:
raise AiException(HttpStatus.SERVER_ERROR, AiErrorCode.YCEA4021007)
return path_flag, cv_image_list
def main():
args = set_default(get_argument())
args, initial_epoch = search_same(args)
if initial_epoch == -1:
# training was already finished!
return
elif initial_epoch == 0:
# first training or training with snapshot
weekday = ["Mon", "Tue", "Wed", "Thu", "Fri", "Sat", "Sun"]
temp = datetime.now()
args.stamp = "{:02d}{:02d}{:02d}_{}_{:02d}_{:02d}_{:02d}".format(
temp.year // 100,
temp.month,
temp.day,
weekday[temp.weekday()],
temp.hour,
temp.minute,
temp.second,
)
get_session(args)
logger = get_logger("MyLogger")
for k, v in vars(args).items():
logger.info("{} : {}".format(k, v))
##########################
# Generator
##########################
trainset, valset, testset = set_dataset(args, logger)
train_generator = dataloader(args, trainset, 'train')
val_generator = dataloader(args, valset, 'val', False)
# for t in train_generator:
# print(t[0]['main_input'].shape, t[0]['main_input'].numpy().min(), t[0]['main_input'].numpy().max(),
# t[1]['main_output'].shape, t[1]['main_output'].numpy().min(), t[1]['main_output'].numpy().max(),
# t[1]['main_output'].numpy().argmax())
# print()
logger.info("TOTAL STEPS OF DATASET FOR TRAINING")
logger.info("========== trainset ==========")
logger.info(" --> {}".format(len(trainset)))
steps_per_epoch = args.steps or len(trainset) // args.batch_size
logger.info(" --> {}".format(steps_per_epoch))
logger.info("=========== valset ===========")
validation_steps = len(valset) // args.batch_size
logger.info(" --> {}".format(len(valset)))
logger.info(" --> {}".format(validation_steps))
##########################
# Model
##########################
model = set_model.Backbone(args, logger)
if args.summary:
model.summary()
return
logger.info("Build model!")
##########################
# Metric
##########################
metrics = {
'loss':
tf.keras.metrics.Mean('loss', dtype=tf.float32),
'acc':
tf.keras.metrics.CategoricalAccuracy('acc', dtype=tf.float32),
'val_loss':
tf.keras.metrics.Mean('val_loss', dtype=tf.float32),
'val_acc':
tf.keras.metrics.CategoricalAccuracy('val_acc', dtype=tf.float32),
}
csvlogger, lr_scheduler = create_callbacks(args, steps_per_epoch, metrics)
if args.optimizer == 'sgd':
optimizer = tf.keras.optimizers.SGD(lr_scheduler,
momentum=.9,
decay=.0001)
elif args.optimizer == 'adam':
optimizer = tf.keras.optimizers.Adam(lr_scheduler)
##########################
# Train
##########################
# steps_per_epoch = 10
# validation_steps = 10
train_iterator = iter(train_generator)
val_iterator = iter(val_generator)
progress_desc_train = 'Train : Loss {:.4f} | Acc {:.4f}'
progress_desc_val = 'Val : Loss {:.4f} | Acc {:.4f}'
for epoch in range(initial_epoch, args.epochs):
print('\nEpoch {}/{}'.format(epoch + 1, args.epochs))
print('Learning Rate : {}'.format(
optimizer.learning_rate(optimizer.iterations)))
progressbar_train = tqdm.tqdm(tf.range(steps_per_epoch),
desc=progress_desc_train.format(
0, 0, 0, 0),
leave=True)
for step in progressbar_train:
inputs = next(train_iterator)
img = inputs[0]['main_input']
label = inputs[1]['main_output']
with tf.GradientTape() as tape:
logits = tf.cast(model(img, training=True), tf.float32)
loss = tf.keras.losses.categorical_crossentropy(label, logits)
loss = tf.reduce_mean(loss)
grads = tape.gradient(loss, model.trainable_variables)
optimizer.apply_gradients(zip(grads, model.trainable_variables))
metrics['loss'].update_state(loss)
metrics['acc'].update_state(label, logits)
progressbar_train.set_description(
progress_desc_train.format(metrics['loss'].result(),
metrics['acc'].result()))
progressbar_train.refresh()
progressbar_val = tqdm.tqdm(tf.range(validation_steps),
desc=progress_desc_val.format(0, 0),
leave=True)
for step in progressbar_val:
val_inputs = next(val_iterator)
val_img = val_inputs[0]['main_input']
val_label = val_inputs[1]['main_output']
val_logits = tf.cast(model(val_img, training=False), tf.float32)
val_loss = tf.keras.losses.categorical_crossentropy(
val_label, val_logits)
val_loss = tf.reduce_mean(val_loss)
metrics['val_loss'].update_state(val_loss)
metrics['val_acc'].update_state(val_label, val_logits)
progressbar_val.set_description(
progress_desc_val.format(metrics['val_loss'].result(),
metrics['val_acc'].result()))
progressbar_val.refresh()
logs = {k: v.result().numpy() for k, v in metrics.items()}
logs['epoch'] = epoch + 1
if args.checkpoint:
model.save_weights(
os.path.join(
args.result_path,
'{}/checkpoint/{:04d}_{:.4f}_{:.4f}.h5'.format(
args.stamp, epoch + 1, logs['val_acc'],
logs['val_loss'])))
print('\nSaved at {}'.format(
os.path.join(
args.result_path,
'{}/checkpoint/{:04d}_{:.4f}_{:.4f}.h5'.format(
args.stamp, epoch + 1, logs['val_acc'],
logs['val_loss']))))
if args.history:
csvlogger = csvlogger.append(logs, ignore_index=True)
csvlogger.to_csv(os.path.join(
args.result_path, '{}/history/epoch.csv'.format(args.stamp)),
index=False)
for k, v in metrics.items():
v.reset_states()
def get(self):
email = get_argument('email')
user = users.user_by_email(email)
return make_response(status=HttpStatus.SUCCESS, data=user)
def post(self):
name = get_argument('username', required=True, help='请输入用户名')
password = get_argument('password', required=False, help='请输入密码')
email = get_argument('email', required=False, help='请输入邮箱')
users.create_user(name, email, password)
return make_response(status=HttpStatus.SUCCESS)
def post(self):
orgName = get_argument("orgName", required=True, help='所属机构不能为空')
deviceName = get_argument("deviceName", required=True, help='所属设备不能为空')
alarmType = get_argument("alarmType", required=True, help='报警类型不能为空')
alarmLevel = get_argument("alarmLevel", required=True, help='报警等级不能为空')
alarmTime = get_argument("alarmTime", required=True, help='报警时间不能为空')
remark = get_argument("remark", required=False)
# 转回img
img = pickle.loads(eval(remark))
# json_data = request.form['json_data']
# dict_data = json.loads(json_data)
# orgName =dict_data.get('orgName',-1)
# deviceName =dict_data.get('deviceName',-1)
# alarmType =dict_data.get('alarmType')
# alarmLevel =dict_data.get('alarmLevel')
# alarmTime =dict_data.get('alarmTime')
# f = request.files['myfile']
# x=f.read()
# print('>>>',x)
# with open('D:/image/b123.jpg', 'ab') as f:
# f.write(x)
# for i in f:
# print('@@',type(i))
# with open('D:/image/a.jpg', 'ab') as f:
# f.write(i)
# 告警类别 00 对应海康 人数超限：'131643', '单人异常', '131644', '多人异常'
# 告警类别 01 对应海康 非法入侵：'131585', '跨域警戒线','131588', '区域入侵', '131586', '人员进入'
if alarmType == '131643':
alarmType = '00'
vlt_err_alrm_inf = '单人异常'
elif alarmType == '131644':
alarmType = '00'
vlt_err_alrm_inf = '多人异常'
elif alarmType == '131585':
alarmType = '01'
vlt_err_alrm_inf = '跨域警戒线'
elif alarmType == '131588':
alarmType = '01'
vlt_err_alrm_inf = '区域入侵'
elif alarmType == '131586':
alarmType = '01'
vlt_err_alrm_inf = '人员进入'
elif alarmType == '150002':
alarmType = '01'
vlt_err_alrm_inf = '非法人员入侵'
else:
raise AiException(HttpStatus.SERVER_ERROR, AiErrorCode.YCEA4021012)
alarm_info_hk_po = AlarmInfoHk()
alarm_info_hk_po.id = gen_uuid()
alarm_info_hk_po.ccbins_id = orgName
alarm_info_hk_po.eqmt_id = deviceName
alarm_info_hk_po.bsn_cgycd = alarmType
alarm_info_hk_po.vlt_err_alrm_inf = vlt_err_alrm_inf
alarm_info_hk_po.alarm_level = alarmLevel
alarm_info_hk_po.stdt_tm = datetime.strptime(alarmTime,
'%Y%m%d %H:%M:%S')
if remark and len(remark) > 0:
try:
cur_date = time.strftime('%Y%m%d', time.localtime(time.time()))
folder_path = yaml_cfg.get(
'illegal_image_path_hk') + "/" + cur_date + '/' + orgName
if not os.path.exists(folder_path):
os.makedirs(folder_path)
img_path = (folder_path + "/" + alarm_info_hk_po.id + ".jpg")
cv.imwrite('D:/image/ab.jpg', img)
# f.save(img_path)
alarm_info_hk_po.uploadfiletrgtrfullnm = img_path
except Exception:
logger.error("存储hk告警图片失败, " + alarmType + "," +
vlt_err_alrm_inf + "," + alarmTime)
alarm_hk_dao.add_record(alarm_info_hk_po)
return make_response(status=HttpStatus.SUCCESS,
data={"code": "success"})
