modify hubserving

deploy/hubserving/.ipynb_checkpoints/serving-checkpoint.md

+# 服务部署
+
+PaddleOCR提供2种服务部署方式：
+- 基于HubServing的部署：已集成到PaddleOCR中（[code](https://github.com/PaddlePaddle/PaddleOCR/tree/develop/deploy/ocr_hubserving)），按照本教程使用；
+- 基于PaddleServing的部署：详见PaddleServing官网[demo](https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/ocr)，后续也将集成到PaddleOCR。  
+
+服务部署目录下包括检测、识别、2阶段串联三种服务包，根据需求选择相应的服务包进行安装和启动。目录如下：
+```
+deploy/hubserving/
+  └─  ocr_det     检测模块服务包
+  └─  ocr_rec     识别模块服务包
+  └─  ocr_system  检测+识别串联服务包
+```
+
+每个服务包下包含3个文件。以2阶段串联服务包为例，目录如下：
+```
+deploy/hubserving/ocr_system/
+  └─  __init__.py    空文件
+  └─  config.json    配置文件，启动服务时作为参数传入
+  └─  module.py      主模块，包含服务的完整逻辑
+```
+
+## 启动服务
+以下步骤以检测+识别2阶段串联服务为例，如果只需要检测服务或识别服务，替换相应文件路径即可。
+### 1. 安装paddlehub
+```pip3 install paddlehub --upgrade -i https://pypi.tuna.tsinghua.edu.cn/simple```
+
+### 2. 安装服务模块
+PaddleOCR提供3种服务模块，根据需要安装所需模块。如： 
+
+安装检测服务模块：  
+```hub install deploy/hubserving/ocr_det/```  
+
+或，安装识别服务模块：  
+```hub install deploy/hubserving/ocr_rec/```  
+
+或，安装检测+识别串联服务模块：  
+```hub install deploy/hubserving/ocr_system/```  
+
+### 3. 修改配置文件
+在config.json中指定模型路径、是否使用GPU、是否对结果做可视化等参数，如，串联服务ocr_system的配置：
+```python
+{
+    "modules_info": {
+        "ocr_system": {
+            "init_args": {
+                "version": "1.0.0",
+                "det_model_dir": "./inference/det/",
+                "rec_model_dir": "./inference/rec/",
+                "use_gpu": true
+            },
+            "predict_args": {
+                "visualization": false
+            }
+        }
+    }
+}
+```
+其中，模型路径对应的模型为```inference模型```。
+
+### 4. 运行启动命令
+```hub serving start -m ocr_system --config hubserving/ocr_det/config.json```  
+
+这样就完成了一个服务化API的部署，默认端口号为8866。
+
+**NOTE:** 如使用GPU预测(即，config中use_gpu置为true)，则需要在启动服务之前，设置CUDA_VISIBLE_DEVICES环境变量，如：```export CUDA_VISIBLE_DEVICES=0```，否则不用设置。
+
+## 发送预测请求
+配置好服务端，以下数行代码即可实现发送预测请求，获取预测结果:
+
+```python
+import requests
+import json
+import cv2
+import base64
+
+def cv2_to_base64(image):
+    return base64.b64encode(image).decode('utf8')
+
+# 发送HTTP请求
+data = {'images':[cv2_to_base64(open("./doc/imgs/11.jpg", 'rb').read())]}
+headers = {"Content-type": "application/json"}
+# url = "http://127.0.0.1:8866/predict/ocr_det"
+# url = "http://127.0.0.1:8866/predict/ocr_rec"
+url = "http://127.0.0.1:8866/predict/ocr_system"
+r = requests.post(url=url, headers=headers, data=json.dumps(data))
+
+# 打印预测结果
+print(r.json()["results"])
+```
+
+你可能需要根据实际情况修改```url```字符串中的端口号和服务模块名称。  
+
+上面所示代码都已写入测试脚本，可直接运行命令：```python tools/test_hubserving.py```
+
+## 自定义修改服务模块
+如果需要修改服务逻辑，你一般需要操作以下步骤：  
+
+1、 停止服务  
+```hub serving stop -m ocr_system```  
+
+2、 到相应的module.py文件中根据实际需求修改代码  
+
+3、 卸载旧服务包  
+```hub uninstall ocr_system```  
+
+4、 安装修改后的新服务包  
+```hub install deploy/hubserving/ocr_system/```  
+

deploy/hubserving/ocr_det/.ipynb_checkpoints/params-checkpoint.py

+# -*- coding:utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+
+class Config(object):
+    pass
+
+
+def read_params():
+    cfg = Config()
+    
+    #params for text detector
+    cfg.det_algorithm = "DB"
+    # cfg.det_model_dir = "./inference/ch_det_mv3_db/"
+    cfg.det_model_dir = "./inference/det/"
+    cfg.det_max_side_len = 960
+
+    #DB parmas
+    cfg.det_db_thresh =0.3
+    cfg.det_db_box_thresh =0.5
+    cfg.det_db_unclip_ratio =2.0
+
+    # #EAST parmas
+    # cfg.det_east_score_thresh = 0.8
+    # cfg.det_east_cover_thresh = 0.1
+    # cfg.det_east_nms_thresh = 0.2
+
+    # #params for text recognizer
+    # cfg.rec_algorithm = "CRNN"
+    # # cfg.rec_model_dir = "./inference/ch_det_mv3_crnn/"
+    # cfg.rec_model_dir = "./inference/rec/"
+
+    # cfg.rec_image_shape = "3, 32, 320"
+    # cfg.rec_char_type = 'ch'
+    # cfg.rec_batch_num = 30
+    # cfg.rec_char_dict_path = "./ppocr/utils/ppocr_keys_v1.txt"
+    # cfg.use_space_char = True
+
+    return cfg
\ No newline at end of file

deploy/ocr_hubserving/ocr_det/config.json → deploy/hubserving/ocr_det/config.json

@@ -3,12 +3,14 @@
         "ocr_det": {
             "init_args": {
                 "version": "1.0.0",
-                "det_model_dir": "./inference/ch_det_mv3_db/",
                 "use_gpu": true
             },
             "predict_args": {
                 "visualization": false
             }
         }
-    }
+    },
+    "port": 8866,
+    "use_multiprocess": false,
+    "workers": 2    
 }

deploy/ocr_hubserving/ocr_det/module.py → deploy/hubserving/ocr_det/module.py

@@ -22,8 +22,6 @@ import paddlehub as hub
 from tools.infer.utility import draw_boxes, base64_to_cv2
 from tools.infer.predict_det import TextDetector
 
-class Config(object):
-    pass
 
 @moduleinfo(
     name="ocr_det",
@@ -33,43 +31,28 @@ class Config(object):
     author_email="paddle-dev@baidu.com",
     type="cv/text_recognition")
 class OCRDet(hub.Module):
-    def _initialize(self, 
-                    det_model_dir="",
-                    det_algorithm="DB",
-                    use_gpu=False
-                    ):
+    def _initialize(self, use_gpu=False):
         """
         initialize with the necessary elements
         """
-        self.config = Config()
-        self.config.use_gpu = use_gpu
+        from ocr_det.params import read_params
+        cfg = read_params()
+
+        cfg.use_gpu = use_gpu
         if use_gpu:
             try:
                 _places = os.environ["CUDA_VISIBLE_DEVICES"]
                 int(_places[0])
                 print("use gpu: ", use_gpu)
                 print("CUDA_VISIBLE_DEVICES: ", _places)
+                cfg.gpu_mem = 8000
             except:
                 raise RuntimeError(
                     "Environment Variable CUDA_VISIBLE_DEVICES is not set correctly. If you wanna use gpu, please set CUDA_VISIBLE_DEVICES via export CUDA_VISIBLE_DEVICES=cuda_device_id."
                 )
-        self.config.ir_optim = True
-        self.config.gpu_mem = 8000
-
-        #params for text detector
-        self.config.det_algorithm = det_algorithm
-        self.config.det_model_dir = det_model_dir
-        # self.config.det_model_dir = "./inference/det/"
-
-        #DB parmas
-        self.config.det_db_thresh =0.3
-        self.config.det_db_box_thresh =0.5
-        self.config.det_db_unclip_ratio =2.0
+        cfg.ir_optim = True
 
-        #EAST parmas
-        self.config.det_east_score_thresh = 0.8
-        self.config.det_east_cover_thresh = 0.1
-        self.config.det_east_nms_thresh = 0.2
+        self.text_detector = TextDetector(cfg)
 
     def read_images(self, paths=[]):
         images = []
@@ -83,10 +66,9 @@ class OCRDet(hub.Module):
             images.append(img)
         return images
 
-    def det_text(self,
+    def predict(self,
                 images=[],
                 paths=[],
-                det_max_side_len=960,
                 draw_img_save='ocr_det_result',
                 visualization=False):
         """
@@ -94,10 +76,8 @@ class OCRDet(hub.Module):
         Args:
             images (list(numpy.ndarray)): images data, shape of each is [H, W, C]. If images not paths
             paths (list[str]): The paths of images. If paths not images
-            use_gpu (bool): Whether to use gpu. Default false.
-            output_dir (str): The directory to store output images.
+            draw_img_save (str): The directory to store output images.
             visualization (bool): Whether to save image or not.
-            box_thresh(float): the threshold of the detected text box's confidence
         Returns:
             res (list): The result of text detection box and save path of images.
         """
@@ -111,8 +91,6 @@ class OCRDet(hub.Module):
 
         assert predicted_data != [], "There is not any image to be predicted. Please check the input data."
         
-        self.config.det_max_side_len = det_max_side_len
-        text_detector = TextDetector(self.config)
         all_results = []
         for img in predicted_data:
             result = {'save_path': ''}
@@ -121,7 +99,7 @@ class OCRDet(hub.Module):
                 result['data'] = []
                 all_results.append(result)
                 continue
-            dt_boxes, elapse = text_detector(img)
+            dt_boxes, elapse = self.text_detector(img)
             print("Predict time : ", elapse)
             result['data'] = dt_boxes.astype(np.int).tolist()
 
@@ -146,7 +124,7 @@ class OCRDet(hub.Module):
         Run as a service.
         """
         images_decode = [base64_to_cv2(image) for image in images]
-        results = self.det_text(images_decode, **kwargs)
+        results = self.predict(images_decode, **kwargs)
         return results
 
    
@@ -156,5 +134,5 @@ if __name__ == '__main__':
         './doc/imgs/11.jpg',
         './doc/imgs/12.jpg',
     ]
-    res = ocr.det_text(paths=image_path, visualization=True)
+    res = ocr.predict(paths=image_path, visualization=True)
     print(res)
\ No newline at end of file

deploy/hubserving/ocr_det/params.py

+# -*- coding:utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+
+class Config(object):
+    pass
+
+
+def read_params():
+    cfg = Config()
+    
+    #params for text detector
+    cfg.det_algorithm = "DB"
+    cfg.det_model_dir = "./inference/ch_det_mv3_db/"
+    cfg.det_max_side_len = 960
+
+    #DB parmas
+    cfg.det_db_thresh =0.3
+    cfg.det_db_box_thresh =0.5
+    cfg.det_db_unclip_ratio =2.0
+
+    # #EAST parmas
+    # cfg.det_east_score_thresh = 0.8
+    # cfg.det_east_cover_thresh = 0.1
+    # cfg.det_east_nms_thresh = 0.2
+
+    # #params for text recognizer
+    # cfg.rec_algorithm = "CRNN"
+    # cfg.rec_model_dir = "./inference/ch_det_mv3_crnn/"
+
+    # cfg.rec_image_shape = "3, 32, 320"
+    # cfg.rec_char_type = 'ch'
+    # cfg.rec_batch_num = 30
+    # cfg.rec_char_dict_path = "./ppocr/utils/ppocr_keys_v1.txt"
+    # cfg.use_space_char = True
+
+    return cfg
\ No newline at end of file

deploy/ocr_hubserving/ocr_rec/config.json → deploy/hubserving/ocr_rec/config.json

@@ -3,11 +3,13 @@
         "ocr_rec": {
             "init_args": {
                 "version": "1.0.0",
-                "det_model_dir": "./inference/ch_rec_mv3_crnn/",
                 "use_gpu": true
             },
             "predict_args": {
             }
         }
-    }
+    },
+    "port": 8867,
+    "use_multiprocess": false,
+    "workers": 2
 }

deploy/ocr_hubserving/ocr_rec/module.py → deploy/hubserving/ocr_rec/module.py

@@ -22,8 +22,6 @@ import paddlehub as hub
 from tools.infer.utility import base64_to_cv2
 from tools.infer.predict_rec import TextRecognizer
 
-class Config(object):
-    pass
 
 @moduleinfo(
     name="ocr_rec",
@@ -33,41 +31,28 @@ class Config(object):
     author_email="paddle-dev@baidu.com",
     type="cv/text_recognition")
 class OCRRec(hub.Module):
-    def _initialize(self, 
-                    rec_model_dir="",
-                    rec_algorithm="CRNN",
-                    rec_char_dict_path="./ppocr/utils/ppocr_keys_v1.txt",
-                    rec_batch_num=30,
-                    use_gpu=False
-                    ):
+    def _initialize(self, use_gpu=False):
         """
         initialize with the necessary elements
         """
-        self.config = Config()
-        self.config.use_gpu = use_gpu
+        from ocr_rec.params import read_params
+        cfg = read_params()
+
+        cfg.use_gpu = use_gpu
         if use_gpu:
             try:
                 _places = os.environ["CUDA_VISIBLE_DEVICES"]
                 int(_places[0])
                 print("use gpu: ", use_gpu)
                 print("CUDA_VISIBLE_DEVICES: ", _places)
+                cfg.gpu_mem = 8000
             except:
                 raise RuntimeError(
                     "Environment Variable CUDA_VISIBLE_DEVICES is not set correctly. If you wanna use gpu, please set CUDA_VISIBLE_DEVICES via export CUDA_VISIBLE_DEVICES=cuda_device_id."
                 )
-        self.config.ir_optim = True
-        self.config.gpu_mem = 8000
+        cfg.ir_optim = True
 
-        #params for text recognizer
-        self.config.rec_algorithm = rec_algorithm
-        self.config.rec_model_dir = rec_model_dir
-        # self.config.rec_model_dir = "./inference/rec/"
-        
-        self.config.rec_image_shape = "3, 32, 320"
-        self.config.rec_char_type = 'ch'
-        self.config.rec_batch_num = rec_batch_num
-        self.config.rec_char_dict_path = rec_char_dict_path
-        self.config.use_space_char = True
+        self.text_recognizer = TextRecognizer(cfg)
 
     def read_images(self, paths=[]):
         images = []
@@ -81,7 +66,7 @@ class OCRRec(hub.Module):
             images.append(img)
         return images
 
-    def rec_text(self,
+    def predict(self,
                 images=[],
                 paths=[]):
         """
@@ -102,14 +87,13 @@ class OCRRec(hub.Module):
 
         assert predicted_data != [], "There is not any image to be predicted. Please check the input data."
         
-        text_recognizer = TextRecognizer(self.config)
         img_list = []
         for img in predicted_data:
             if img is None:
                 continue
             img_list.append(img)
         try:
-            rec_res, predict_time = text_recognizer(img_list)
+            rec_res, predict_time = self.text_recognizer(img_list)
         except Exception as e:
             print(e)
             return []
@@ -121,7 +105,7 @@ class OCRRec(hub.Module):
         Run as a service.
         """
         images_decode = [base64_to_cv2(image) for image in images]
-        results = self.det_text(images_decode, **kwargs)
+        results = self.predict(images_decode, **kwargs)
         return results
 
    
@@ -132,5 +116,5 @@ if __name__ == '__main__':
         './doc/imgs_words/ch/word_2.jpg',
         './doc/imgs_words/ch/word_3.jpg',
     ]
-    res = ocr.rec_text(paths=image_path)
+    res = ocr.predict(paths=image_path)
     print(res)
\ No newline at end of file

deploy/hubserving/ocr_rec/params.py

+# -*- coding:utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+
+class Config(object):
+    pass
+
+
+def read_params():
+    cfg = Config()
+    
+    # #params for text detector
+    # cfg.det_algorithm = "DB"
+    # cfg.det_model_dir = "./inference/ch_det_mv3_db/"
+    # cfg.det_max_side_len = 960
+
+    # #DB parmas
+    # cfg.det_db_thresh =0.3
+    # cfg.det_db_box_thresh =0.5
+    # cfg.det_db_unclip_ratio =2.0
+
+    # #EAST parmas
+    # cfg.det_east_score_thresh = 0.8
+    # cfg.det_east_cover_thresh = 0.1
+    # cfg.det_east_nms_thresh = 0.2
+
+    #params for text recognizer
+    cfg.rec_algorithm = "CRNN"
+    cfg.rec_model_dir = "./inference/ch_rec_mv3_crnn/"
+
+    cfg.rec_image_shape = "3, 32, 320"
+    cfg.rec_char_type = 'ch'
+    cfg.rec_batch_num = 30
+    cfg.rec_char_dict_path = "./ppocr/utils/ppocr_keys_v1.txt"
+    cfg.use_space_char = True
+
+    return cfg
\ No newline at end of file

deploy/ocr_hubserving/ocr_system/config.json → deploy/hubserving/ocr_system/config.json

@@ -3,14 +3,15 @@
         "ocr_system": {
             "init_args": {
                 "version": "1.0.0",
-                "det_model_dir": "./inference/ch_det_mv3_db/",
-                "rec_model_dir": "./inference/ch_rec_mv3_crnn/",
                 "use_gpu": true
             },
             "predict_args": {
                 "visualization": false
             }
         }
-    }
+    },
+    "port": 8868,
+    "use_multiprocess": false,
+    "workers": 2
 }
 

deploy/ocr_hubserving/ocr_system/module.py → deploy/hubserving/ocr_system/module.py

@@ -23,9 +23,6 @@ from tools.infer.utility import draw_ocr, base64_to_cv2
 from tools.infer.predict_system import TextSystem
 
 
-class Config(object):
-    pass
-
 @moduleinfo(
     name="ocr_system",
     version="1.0.0",
@@ -34,58 +31,28 @@ class Config(object):
     author_email="paddle-dev@baidu.com",
     type="cv/text_recognition")
 class OCRSystem(hub.Module):
-    def _initialize(self, 
-                    det_model_dir="",
-                    det_algorithm="DB",
-                    rec_model_dir="",
-                    rec_algorithm="CRNN",
-                    rec_char_dict_path="./ppocr/utils/ppocr_keys_v1.txt",
-                    rec_batch_num=30,
-                    use_gpu=False
-                    ):
+    def _initialize(self, use_gpu=False):
         """
         initialize with the necessary elements
         """
-        self.config = Config()
-        self.config.use_gpu = use_gpu
+        from ocr_system.params import read_params
+        cfg = read_params()
+
+        cfg.use_gpu = use_gpu
         if use_gpu:
             try:
                 _places = os.environ["CUDA_VISIBLE_DEVICES"]
                 int(_places[0])
                 print("use gpu: ", use_gpu)
                 print("CUDA_VISIBLE_DEVICES: ", _places)
+                cfg.gpu_mem = 8000
             except:
                 raise RuntimeError(
                     "Environment Variable CUDA_VISIBLE_DEVICES is not set correctly. If you wanna use gpu, please set CUDA_VISIBLE_DEVICES via export CUDA_VISIBLE_DEVICES=cuda_device_id."
                 )
-        self.config.ir_optim = True
-        self.config.gpu_mem = 8000
-
-        #params for text detector
-        self.config.det_algorithm = det_algorithm
-        self.config.det_model_dir = det_model_dir
-        # self.config.det_model_dir = "./inference/det/"
-
-        #DB parmas
-        self.config.det_db_thresh =0.3
-        self.config.det_db_box_thresh =0.5
-        self.config.det_db_unclip_ratio =2.0
-
-        #EAST parmas
-        self.config.det_east_score_thresh = 0.8
-        self.config.det_east_cover_thresh = 0.1
-        self.config.det_east_nms_thresh = 0.2
-
-        #params for text recognizer
-        self.config.rec_algorithm = rec_algorithm
-        self.config.rec_model_dir = rec_model_dir
-        # self.config.rec_model_dir = "./inference/rec/"
-
-        self.config.rec_image_shape = "3, 32, 320"
-        self.config.rec_char_type = 'ch'
-        self.config.rec_batch_num = rec_batch_num
-        self.config.rec_char_dict_path = rec_char_dict_path
-        self.config.use_space_char = True
+        cfg.ir_optim = True
+
+        self.text_sys = TextSystem(cfg)
 
     def read_images(self, paths=[]):
         images = []
@@ -99,10 +66,9 @@ class OCRSystem(hub.Module):
             images.append(img)
         return images
 
-    def recognize_text(self,
+    def predict(self,
                        images=[],
                        paths=[],
-                       det_max_side_len=960,
                        draw_img_save='ocr_result',
                        visualization=False,
                        text_thresh=0.5):
@@ -111,11 +77,8 @@ class OCRSystem(hub.Module):
         Args:
             images (list(numpy.ndarray)): images data, shape of each is [H, W, C]. If images not paths
             paths (list[str]): The paths of images. If paths not images
-            use_gpu (bool): Whether to use gpu.
-            batch_size(int): the program deals once with one
-            output_dir (str): The directory to store output images.
+            draw_img_save (str): The directory to store output images.
             visualization (bool): Whether to save image or not.
-            box_thresh(float): the threshold of the detected text box's confidence
             text_thresh(float): the threshold of the recognize chinese texts' confidence
         Returns:
             res (list): The result of chinese texts and save path of images.
@@ -130,8 +93,6 @@ class OCRSystem(hub.Module):
 
         assert predicted_data != [], "There is not any image to be predicted. Please check the input data."
 
-        self.config.det_max_side_len = det_max_side_len
-        text_sys = TextSystem(self.config)
         cnt = 0
         all_results = []
         for img in predicted_data:
@@ -142,7 +103,7 @@ class OCRSystem(hub.Module):
                 all_results.append(result)
                 continue
             starttime = time.time()
-            dt_boxes, rec_res = text_sys(img)
+            dt_boxes, rec_res = self.text_sys(img)
             elapse = time.time() - starttime
             cnt += 1
             print("Predict time of image %d: %.3fs" % (cnt, elapse))
@@ -187,7 +148,7 @@ class OCRSystem(hub.Module):
         Run as a service.
         """
         images_decode = [base64_to_cv2(image) for image in images]
-        results = self.recognize_text(images_decode, **kwargs)
+        results = self.predict(images_decode, **kwargs)
         return results
 
    
@@ -197,5 +158,5 @@ if __name__ == '__main__':
         './doc/imgs/11.jpg',
         './doc/imgs/12.jpg',
     ]
-    res = ocr.recognize_text(paths=image_path, visualization=True)
+    res = ocr.predict(paths=image_path, visualization=False)
     print(res)
\ No newline at end of file

deploy/hubserving/ocr_system/params.py

+# -*- coding:utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+
+class Config(object):
+    pass
+
+
+def read_params():
+    cfg = Config()
+    
+    #params for text detector
+    cfg.det_algorithm = "DB"
+    cfg.det_model_dir = "./inference/ch_det_mv3_db/"
+    cfg.det_max_side_len = 960
+
+    #DB parmas
+    cfg.det_db_thresh =0.3
+    cfg.det_db_box_thresh =0.5
+    cfg.det_db_unclip_ratio =2.0
+
+    #EAST parmas
+    cfg.det_east_score_thresh = 0.8
+    cfg.det_east_cover_thresh = 0.1
+    cfg.det_east_nms_thresh = 0.2
+
+    #params for text recognizer
+    cfg.rec_algorithm = "CRNN"
+    cfg.rec_model_dir = "./inference/ch_rec_mv3_crnn/"
+
+    cfg.rec_image_shape = "3, 32, 320"
+    cfg.rec_char_type = 'ch'
+    cfg.rec_batch_num = 30
+    cfg.rec_char_dict_path = "./ppocr/utils/ppocr_keys_v1.txt"
+    cfg.use_space_char = True
+
+    return cfg
\ No newline at end of file

doc/doc_ch/serving.md

 # 服务部署
 
 PaddleOCR提供2种服务部署方式：
-- 基于HubServing的部署：已集成到PaddleOCR中（[code](https://github.com/PaddlePaddle/PaddleOCR/tree/develop/deploy/ocr_hubserving)），按照本教程使用；
+- 基于HubServing的部署：已集成到PaddleOCR中（[code](https://github.com/PaddlePaddle/PaddleOCR/tree/develop/deploy/hubserving)），按照本教程使用；
 - 基于PaddleServing的部署：详见PaddleServing官网[demo](https://github.com/PaddlePaddle/Serving/tree/develop/python/examples/ocr)，后续也将集成到PaddleOCR。  
 
 服务部署目录下包括检测、识别、2阶段串联三种服务包，根据需求选择相应的服务包进行安装和启动。目录如下：
@@ -15,12 +15,13 @@ deploy/hubserving/
 每个服务包下包含3个文件。以2阶段串联服务包为例，目录如下：
 ```
 deploy/hubserving/ocr_system/
-  └─  __init__.py    空文件
-  └─  config.json    配置文件，启动服务时作为参数传入
-  └─  module.py      主模块，包含服务的完整逻辑
+  └─  __init__.py    空文件，必选
+  └─  config.json    配置文件，可选，使用配置启动服务时作为参数传入
+  └─  module.py      主模块，必选，包含服务的完整逻辑
+  └─  params.py      参数文件，必选，包含模型路径、前后处理参数等参数
 ```
 
-## 启动服务
+## 快速启动服务
 以下步骤以检测+识别2阶段串联服务为例，如果只需要检测服务或识别服务，替换相应文件路径即可。
 ### 1. 安装paddlehub
 ```pip3 install paddlehub --upgrade -i https://pypi.tuna.tsinghua.edu.cn/simple```
@@ -31,39 +32,71 @@ PaddleOCR提供3种服务模块，根据需要安装所需模块。如：
 安装检测服务模块：  
 ```hub install deploy/hubserving/ocr_det/```  
 
-或，安装识别服务模块：  
+或，安装识别服务模块：    
 ```hub install deploy/hubserving/ocr_rec/```  
 
 或，安装检测+识别串联服务模块：  
 ```hub install deploy/hubserving/ocr_system/```  
 
-### 3. 修改配置文件
-在config.json中指定模型路径、是否使用GPU、是否对结果做可视化等参数，如，串联服务ocr_system的配置：
+### 3. 启动服务
+#### 方式1. 命令行命令启动（仅支持CPU）
+**启动命令：**  
+```shell
+$ hub serving start --modules [Module1==Version1, Module2==Version2, ...] \
+                    --port XXXX \
+                    --use_multiprocess \
+                    --workers \
+```  
+
+**参数：**  
+
+|参数|用途|  
+|-|-|  
+|--modules/-m|PaddleHub Serving预安装模型，以多个Module==Version键值对的形式列出<br>*`当不指定Version时，默认选择最新版本`*|  
+|--port/-p|服务端口，默认为8866|  
+|--use_multiprocess|是否启用并发方式，默认为单进程方式，推荐多核CPU机器使用此方式<br>*`Windows操作系统只支持单进程方式`*|
+|--workers|在并发方式下指定的并发任务数，默认为`2*cpu_count-1`，其中`cpu_count`为CPU核数|  
+
+如启动串联服务：  ```hub serving start -m ocr_system```  
+
+这样就完成了一个服务化API的部署，使用默认端口号8866。
+
+#### 方式2. 配置文件启动（支持CPU、GPU）
+**启动命令：**  
+```hub serving start --config/-c config.json```  
+
+其中，`config.json`格式如下： 
 ```python
 {
     "modules_info": {
         "ocr_system": {
             "init_args": {
                 "version": "1.0.0",
-                "det_model_dir": "./inference/det/",
-                "rec_model_dir": "./inference/rec/",
                 "use_gpu": true
             },
             "predict_args": {
                 "visualization": false
             }
         }
-    }
+    },
+    "port": 8868,
+    "use_multiprocess": false,
+    "workers": 2
 }
 ```
-其中，模型路径对应的模型为```inference模型```。
 
-### 4. 运行启动命令
-```hub serving start -m ocr_system --config hubserving/ocr_det/config.json```  
+- `init_args`中的可配参数与`module.py`中的`_initialize`函数接口一致。其中，**当`use_gpu`为`true`时，表示使用GPU启动服务**。  
+- `predict_args`中的可配参数与`module.py`中的`predict`函数接口一致。
 
-这样就完成了一个服务化API的部署，默认端口号为8866。
+**注意:**  
+- 使用配置文件启动服务时，其他参数会被忽略。
+- 如果使用GPU预测(即，`use_gpu`置为`true`)，则需要在启动服务之前，设置CUDA_VISIBLE_DEVICES环境变量，如：```export CUDA_VISIBLE_DEVICES=0```，否则不用设置。
 
-**NOTE:** 如使用GPU预测(即，config中use_gpu置为true)，则需要在启动服务之前，设置CUDA_VISIBLE_DEVICES环境变量，如：```export CUDA_VISIBLE_DEVICES=0```，否则不用设置。
+如，使用GPU 3号卡启动串联服务：  
+```shell
+export CUDA_VISIBLE_DEVICES=3
+hub serving start -c deploy/hubserving/ocr_system/config.json
+```  
 
 ## 发送预测请求
 配置好服务端，以下数行代码即可实现发送预测请求，获取预测结果:
@@ -89,21 +122,25 @@ r = requests.post(url=url, headers=headers, data=json.dumps(data))
 print(r.json()["results"])
 ```
 
-你可能需要根据实际情况修改```url```字符串中的端口号和服务模块名称。  
+你可能需要根据实际情况修改`url`字符串中的端口号和服务模块名称。  
 
 上面所示代码都已写入测试脚本，可直接运行命令：```python tools/test_hubserving.py```
 
 ## 自定义修改服务模块
-如果需要修改服务逻辑，你一般需要操作以下步骤：  
+如果需要修改服务逻辑，你一般需要操作以下步骤（以修改`ocr_system`为例）：  
 
-1、 停止服务  
-```hub serving stop -m ocr_system```  
+- 1、 停止服务  
+```hub serving stop --port/-p XXXX```  
 
-2、 到相应的module.py文件中根据实际需求修改代码  
+- 2、 到相应的`module.py`和`params.py`等文件中根据实际需求修改代码。  
+例如，如果需要替换部署服务所用模型，则需要到`params.py`中修改模型路径参数`det_model_dir`和`rec_model_dir`，当然，同时可能还需要修改其他相关参数，请根据实际情况修改调试。 建议修改后先直接运行`module.py`调试，能正确运行预测后再启动服务测试。
 
-3、 卸载旧服务包  
+- 3、 卸载旧服务包  
 ```hub uninstall ocr_system```  
 
-4、 安装修改后的新服务包  
+- 4、 安装修改后的新服务包  
 ```hub install deploy/hubserving/ocr_system/```  
 
+- 5、重新启动服务  
+```hub serving start -m ocr_system```  
+