算法交互模型去除多线程加载

concurrency/IntelligentRecognitionProcess.py

 # -*- coding: utf-8 -*-
+import json
 import os
 import time
 import copy
         LogUtils.init_log(self.content)
         # 程序开始时间
         start_time = time.time()
-        mod_thread = Common(None, func=get_model, args=(self.config, str(self.gpu_ids[0]), self.msg["models"]))
-        mod_thread.setDaemon(True)
-        mod_thread.start()
-        mod = None
-        model_type_code = None
-        modelConfig = None
+        mod, model_type_code, modelConfig = get_model((self.config, str(self.gpu_ids[0]), self.msg["models"]))
+        # mod_thread = Common(None, func=get_model, args=(self.config, str(self.gpu_ids[0]), self.msg["models"]))
+        # mod_thread.setDaemon(True)
+        # mod_thread.start()
+        # mod = None
+        # model_type_code = None
+        # modelConfig = None
         # 启动公共进程包含（图片上传线程，心跳线程，问题反馈线程）
         commonProcess = CommonProcess(self.fbQueue, None, self.content, self.msg, self.imageQueue,
                                       AnalysisType.ONLINE.value)
             pull_start_time = None
             read_start_time = None
             # 模型初始化次数
-            model = 0
+            # model = 0
             while True:
                 end_time = time.time()
                 create_task_time = end_time - start_time
                     cv2tool.build_cv2()
                     continue
                 read_start_time = None
-                if mod is None and model == 0:
-                    model += 1
-                    logger.info("初始化模型: {}次, requestId: {}", model, self.msg.get("request_id"))
-                    mod, model_type_code, modelConfig = mod_thread.get_result()
+                # if mod is None and model == 0:
+                #     model += 1
+                #     logger.info("初始化模型: {}次, requestId: {}", model, self.msg.get("request_id"))
+                #     mod, model_type_code, modelConfig = mod_thread.get_result()
                 # time00 = time.time()
                 # 调用AI模型
                 p_result, timeOut = mod.process(copy.deepcopy(frame), modelConfig)
         LogUtils.init_log(self.content)
         # 程序开始时间
         start_time = time.time()
-        mod_thread = Common(None, func=get_model, args=(self.config, str(self.gpu_ids[0]), self.msg["models"]))
-        mod_thread.setDaemon(True)
-        mod_thread.start()
-        mod = None
-        model_type_code = None
-        modelConfig = None
+        mod, model_type_code, modelConfig = get_model((self.config, str(self.gpu_ids[0]), self.msg["models"]))
+        # mod_thread = Common(None, func=get_model, args=(self.config, str(self.gpu_ids[0]), self.msg["models"]))
+        # mod_thread.setDaemon(True)
+        # mod_thread.start()
+        # mod = None
+        # model_type_code = None
+        # modelConfig = None
         # 创建心跳队列
         hbQueue = Queue()
         # 结果反馈进程启动
             high_score_image = {}
             step = int(self.content["service"]["frame_step"])
             # 模型初始化速度
-            model = 0
+            # model = 0
             # 总视频帧数
             all_f = None
             if cv2tool.cap is not None:
                     logger.info("任务开始结束分析, requestId: {}", self.msg.get("request_id"))
                     self.stop_task(cv2tool, aiFilePath, AnalysisStatus.SUCCESS.value)
                     break
-                if mod is None and model == 0:
-                    model += 1
-                    logger.info("初始化模型: {}次, requestId: {}", model, self.msg.get("request_id"))
-                    mod, model_type_code, modelConfig = mod_thread.get_result()
+                # if mod is None and model == 0:
+                #     model += 1
+                #     logger.info("初始化模型: {}次, requestId: {}", model, self.msg.get("request_id"))
+                #     mod, model_type_code, modelConfig = mod_thread.get_result()
                 # time00 = time.time()
                 # 调用AI模型
                 p_result, timeOut = mod.process(copy.deepcopy(frame), modelConfig)
 
 
 def get_model(args):
+    logger.info("######################开始加载模型######################")
     for model in args[2]:
         try:
             code = model.get("code")
             needed_objectsIndex = [int(category.get("id")) for category in model.get("categories")]
             logger.info("code:{}, 检查目标：{}, gpuId:{}", code, needed_objectsIndex, args[1])
             if code == ModelType.WATER_SURFACE_MODEL.value[1]:
-                return ModelUtils.SZModel(args[1], needed_objectsIndex), code, args[0].get("sz")
+                logger.info("######################加载河道模型######################")
+                mod, model_type_code, modelConfig = ModelUtils.SZModel(args[1], needed_objectsIndex), code, args[0].get("sz")
+                return mod, model_type_code, modelConfig
             elif code == ModelType.FOREST_FARM_MODEL.value[1]:
+                logger.info("######################加载林场模型######################")
                 return ModelUtils.LCModel(args[1], needed_objectsIndex), code, args[0].get("lc")
             else:
                 logger.error("未匹配到对应的模型")
                 raise ServiceException(ExceptionType.AI_MODEL_MATCH_EXCEPTION.value[0],
                                        ExceptionType.AI_MODEL_MATCH_EXCEPTION.value[1])
         except Exception as e:
-            logger.error("获取模型配置异常：")
-            logger.exception(e)
+            logger.exception("获取模型配置异常: {}", e)
             raise ServiceException(ExceptionType.AI_MODEL_CONFIG_EXCEPTION.value[0],
                                    ExceptionType.AI_MODEL_CONFIG_EXCEPTION.value[1])