CN110989995A - Processing method and system for artificial intelligence application based on open source deep learning framework - Google Patents
Processing method and system for artificial intelligence application based on open source deep learning framework Download PDFInfo
- Publication number
- CN110989995A CN110989995A CN201911182343.3A CN201911182343A CN110989995A CN 110989995 A CN110989995 A CN 110989995A CN 201911182343 A CN201911182343 A CN 201911182343A CN 110989995 A CN110989995 A CN 110989995A
- Authority
- CN
- China
- Prior art keywords
- framework
- artificial intelligence
- interface function
- intelligence application
- deep learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/30—Creation or generation of source code
- G06F8/31—Programming languages or programming paradigms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/40—Transformation of program code
- G06F8/41—Compilation
- G06F8/44—Encoding
- G06F8/447—Target code generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/70—Software maintenance or management
- G06F8/76—Adapting program code to run in a different environment; Porting
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Stored Programmes (AREA)
Abstract
The invention discloses a processing method and a system of artificial intelligence application based on an open source deep learning framework, and relates to two open source deep learning frameworks which are respectively a first framework and a second framework, wherein the artificial intelligence application is developed based on the first framework; the processing method comprises the following steps: acquiring an interface function in a first frame used in a source code of artificial intelligence application, wherein the interface function is a first type of interface function; the function of each first-class interface function is realized again based on the API provided by the second framework so as to obtain a second-class interface function corresponding to each first-class interface function; the second type interface function has the same interface with the corresponding first type interface function; compiling all second type interface functions based on the API provided by the second framework to generate a packaging library; and recompiling the source code of the artificial intelligence application based on the packaging library and the API provided by the second framework so as to generate the artificial intelligence application running on the second framework. The invention can save the development time of the upper-layer AI application and improve the transplantation efficiency.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a processing method and a system for artificial intelligence application based on an open source deep learning framework.
Background
The Caffe (Convolutional neural network framework) is a very widely used open source deep learning framework. One of the significant advantages is that the method has a large number of well-trained classical models, such as AlexNet, VGG, inclusion and the like, and also comprises high-level models, such as ResNet and the like. These models are collected in its Model Zoo and can be used conveniently by developers of AI (artificial intelligence) applications. Because Caffe is highly well known, it is widely used in leading-edge industries and academia. Many papers that provide deep learning of source codes use Caffe to implement their model. That is, a large number of models and AI applications based on the Caffe framework are currently reserved. These AI applications cover various fields of artificial intelligence such as image recognition, face recognition, object detection, speech recognition, etc.
With the rapid development of the artificial intelligence industry in recent years, various new deep learning frameworks are continuously emerging. These frameworks are all characterized, and some excellent deep learning frameworks, regardless of the application range of the hardware platform, the operation efficiency and the power consumption, exceed the Caffe framework. Therefore, more and more AI applications hope to shift from the original Caffe framework to these more excellent deep learning frameworks, and to continuously improve the user experience by achieving higher accuracy and performance index. Therefore, how to rapidly migrate the model and AI application based on the Caffe framework to other deep learning frameworks becomes a hot topic.
Because Caffe is incompatible with other deep learning frameworks, the Caffe-based AI applications can only run on the Caffe framework, and they cannot work normally when the deep learning framework is changed. In the process of transplanting the Caffe-based AI application to other deep learning frameworks, a great deal of modification, sometimes even re-development, needs to be performed on the code of the upper-layer AI application, so that the workload is huge and the transplanting efficiency is low.
Disclosure of Invention
The invention aims to overcome the defects of huge workload and low transplanting efficiency in the process of transplanting an AI application based on an open source deep learning frame to other deep learning frames in the prior art, and provides a processing method and a system for artificial intelligence application based on the open source deep learning frame, which can directly compile and run in other deep learning frame environments without modifying the AI application based on the original open source deep learning frame, thereby greatly reducing the transplanting workload and improving the transplanting efficiency.
The invention solves the technical problems through the following technical scheme:
the invention provides a processing method of artificial intelligence application based on an open source deep learning framework, which relates to two open source deep learning frameworks which are respectively a first framework and a second framework, wherein the artificial intelligence application is developed based on the first framework;
the processing method comprises the following steps:
acquiring all interface functions used in the first frame in the source code of the artificial intelligence application, wherein the interface functions are first-class interface functions;
based on an Application Programming Interface (APII) provided by the second framework, re-realizing the function of each first-class Interface function to obtain a second-class Interface function corresponding to each first-class Interface function;
the interfaces of the second type interface function and the corresponding first type interface function are the same;
compiling all the second type interface functions based on the API provided by the second framework to generate a packaging library;
recompiling the source code of the artificial intelligence application based on the packaging library and the API provided by the second framework so as to generate the artificial intelligence application running on the second framework.
Preferably, the processing method further comprises the following steps:
acquiring all data definitions in the first framework used in source codes of the artificial intelligence application;
packaging all the data definitions into the packaging library.
Preferably, the first framework is Caffe;
the first type of interface function comprises at least one of the following functions:
net- > Forward, Net- > copyTrainedLayerFrom, Net- > reset, Net- > input _ blob () and Net- > output _ blob ().
Preferably, the second framework is Tengine or Ncnn.
Preferably, the processing method further comprises the following steps:
acquiring basic data used in a source code of the artificial intelligence application, wherein the basic data comprises a header file name, a data name, a name of the first type of interface function, a parameter type of the first type of interface function and the number of the first type of interface function;
packaging all the basic data into the packaging library.
The invention also provides a processing system of artificial intelligence application based on the open source deep learning framework, wherein the processing system relates to two open source deep learning frameworks which are respectively a first framework and a second framework, and the artificial intelligence application is developed based on the first framework;
the processing system comprises:
an obtaining module, configured to obtain all interface functions used in the source code of the artificial intelligence application in the first framework, where the interface functions are first-class interface functions;
the realization module is used for realizing the function of each first-class interface function again based on the API provided by the second framework so as to obtain a second-class interface function corresponding to each first-class interface function; the interfaces of the second type interface function and the corresponding first type interface function are the same;
the packaging module is used for compiling all the second type interface functions based on the API provided by the second framework so as to generate a packaging library;
and the compiling module is used for recompiling the source code of the artificial intelligence application based on the packaging library and the API provided by the second framework so as to generate the artificial intelligence application running on the second framework.
Preferably, the obtaining module is further configured to obtain all data definitions in the first framework used in the source code of the artificial intelligence application;
the packaging module is also used for packaging all the data definitions into the packaging library.
Preferably, the first framework is Caffe;
the first type of interface function comprises at least one of the following functions:
net- > Forward, Net- > copyTrainedLayerFrom, Net- > reset, Net- > input _ blob () and Net- > output _ blob ().
Preferably, the second framework is Tengine or Ncnn.
Preferably, the obtaining module is further configured to obtain basic data used in the source code of the artificial intelligence application, where the basic data includes a header file name, a data name, a name of the first type of interface function, a parameter type of the first type of interface function, and a number of the first type of interface function;
the packaging module is also used for packaging all the basic data into the packaging library.
The positive progress effects of the invention are as follows:
according to the invention, by designing and constructing the packaging library conforming to the interface of the original open source deep learning framework, the AI application program originally using the first framework can be compiled and run directly under other deep learning framework environments without any modification, thereby improving the flexibility of the AI application. Further, the invention realizes the transplantation of the Caffe framework to other deep learning frameworks. Because the number of interfaces externally provided by the Caffe framework is limited, the workload can be greatly reduced compared with the process of re-developing the upper-layer AI application for encapsulating and re-realizing the interface function, and the method has certain universality, so that the development time and labor cost of the upper-layer AI application can be greatly saved, and the transplantation efficiency is improved.
Drawings
Fig. 1 is a flowchart of a processing method of artificial intelligence application based on an open source deep learning framework according to embodiment 1 of the present invention.
Fig. 2 is a schematic block diagram of a processing system for artificial intelligence application based on an open source deep learning framework according to embodiment 2 of the present invention.
Detailed Description
The invention is further illustrated by the following examples, which are not intended to limit the scope of the invention.
Example 1
The embodiment provides a processing method of artificial intelligence application based on an open source deep learning framework, the processing method relates to two open source deep learning frameworks which are respectively a first framework and a second framework, and AI application is developed based on the first framework. In this embodiment, the first frame is Caffe, and the second frame is Tengine.
As shown in fig. 1, the processing method provided in this embodiment includes the following steps:
It should be noted that, in the summarizing process, the data definition format and the definition form of the basic data cannot be changed, the basic data includes a header file name, a data name, a name of the first type interface function, a parameter type of the first type interface function, the number of the first type interface function, and the like, and all of these contents are consistent with the original contents.
In this embodiment, the related first interface functions include the following functions: net- > Forward, Net- > copytrainedLayerFrom, Net- > reset, Net- > input _ blob () and Net- > output _ blob (), which are all functions supported by the API provided by the Caffe framework, and the usage and functions are not described herein again.
And 102, re-realizing the function of each first-class interface function based on the API provided by the second framework to obtain a second-class interface function corresponding to each first-class interface function.
The second type interface function has the same interface with the corresponding first type interface function. It should be noted that, to implement the function of each first-type interface function, a use method of each first-type interface function and a detailed function implemented inside need to be analyzed in advance, and a specific implementation manner can be known by those skilled in the art, which is not described herein again.
In this step, the API interface provided by the second framework is used to implement the function of each first interface function again, and in the implementation process, it is ensured that the functions of the second interface function, the input and output of data, the influence on global data, and the like are consistent with the corresponding first interface function. Therefore, when the source code of the AI application is linked to the second type interface function in the subsequent packaging library, the source code can be ensured to be in seamless connection as the source code originally linked to the API of Caffe.
In this step, the second type interface function defined and realized again by the related data is compiled into a packaging library of the Caffe interface, that is, the interface of the second type interface function in the packaging library is completely the same as the interface of the first type interface function provided by Caffe.
And 104, recompiling the source code of the AI application based on the encapsulation library and the API provided by the second framework to generate the AI application capable of running on the second framework.
In the step, the AI application based on Caffe can be run on a new deep learning framework by recompiling and linking the packaging library.
It should be noted that the second framework in the present invention may also be other open source deep learning frameworks, such as Ncnn.
The processing method provided by this embodiment includes a step of encapsulating external data and interface functions of the Caffe framework, and encapsulates the external data and interface functions of Caffe used in the AI application into a single encapsulation library. The data and the interface function in the packaging library are completely consistent with those provided by the Caffe framework in the external use form, but the internal implementation of the interface function is completely different according to the deep learning framework which is operated according to actual needs.
The embodiment provides a method for rapidly accelerating the transplantation of the Caffe-based AI application, so that the AI application program originally using Caffe can be compiled and run directly in Tengine environment without any modification, thereby improving the flexibility of the AI application. Because the number of interfaces externally provided by the Caffe framework is limited, the workload can be greatly reduced compared with the process of re-developing the upper-layer AI application for encapsulating and re-realizing the interface function, and the method has certain universality, so that the development time and labor cost of the upper-layer AI application can be greatly saved, and the transplantation efficiency is improved.
Example 2
As shown in fig. 2, the present embodiment provides a processing system for an artificial intelligence application based on an open source deep learning framework, the processing system relates to two open source deep learning frameworks, namely a first framework and a second framework, respectively, and an AI application is developed based on the first framework. In this embodiment, the first frame is Caffe, and the second frame is Ncnn. The processing system provided by the embodiment includes an obtaining module 1, an implementing module 2, a packaging module 3, and a compiling module 4.
The obtaining module 1 is configured to obtain, through collection and aggregation, basic data used in a source code of an AI application, all data definitions in a Caffe framework, and all interface functions, where the interface functions are first-class interface functions.
It should be noted that, in the summarizing process, the data definition format and the definition form of the basic data cannot be changed, the basic data includes a header file name, a data name, a name of the first type interface function, a parameter type of the first type interface function, the number of the first type interface function, and the like, and all of these contents are consistent with the original contents.
In this embodiment, the related first interface functions include the following functions: net- > Forward, Net- > copyTrainedLayerFrom, Net- > reset, Net- > input _ blob () and Net- > output _ blob ().
And the implementation module 2 is configured to implement the function of each first-class interface function again based on the API provided by the second framework, so as to obtain a second-class interface function corresponding to each first-class interface function. The second type interface function has the same interface with the corresponding first type interface function. The realization of the function of each first-class interface function requires the analysis of the use method of each first-class interface function in advance and the detailed function realized inside.
The implementation module 2 uses the API interface provided by the second framework to implement its function again for each first-class interface function, and in the implementation process, it is ensured that the functions of the second-class interface function, the input and output of data, the influence on global data, and the like are consistent with the corresponding first-class interface function.
The packaging module 3 is used for compiling all the second type interface functions based on the API provided by the second framework so as to generate a packaging library; namely, the related data definition and the second type of re-implemented interface function are compiled into a packaging library of the Caffe interface.
And the compiling module 4 is used for recompiling the source code of the AI application based on the packaging library and the API provided by the second framework so as to generate the AI application running on the second framework.
The compiling module 4 can run on the new deep learning framework by recompiling and linking the packaging library based on the AI application of Caffe.
It should be noted that the second framework in the present invention may also be other open source deep learning frameworks, such as Tengine.
The embodiment provides a transplantation system for rapidly accelerating the Caffe-based AI application, so that the AI application program originally using Caffe can be compiled and run directly in Tengine environment without any modification, thereby improving the flexibility of the AI application. Because the number of interfaces externally provided by the Caffe framework is limited, the workload can be greatly reduced compared with the process of re-developing the upper-layer AI application for encapsulating and re-realizing the interface function, and the method has certain universality, so that the development time and labor cost of the upper-layer AI application can be greatly saved, and the transplantation efficiency is improved.
While specific embodiments of the invention have been described above, it will be appreciated by those skilled in the art that this is by way of example only, and that the scope of the invention is defined by the appended claims. Various changes and modifications to these embodiments may be made by those skilled in the art without departing from the spirit and scope of the invention, and these changes and modifications are within the scope of the invention.
Claims (10)
1. A processing method of artificial intelligence application based on an open source deep learning framework relates to two open source deep learning frameworks which are a first framework and a second framework respectively, wherein the artificial intelligence application is developed based on the first framework;
the processing method is characterized by comprising the following steps:
acquiring all interface functions used in the first frame in the source code of the artificial intelligence application, wherein the interface functions are first-class interface functions;
based on the API provided by the second framework, the function of each first-class interface function is realized again, so that a second-class interface function corresponding to each first-class interface function is obtained;
the interfaces of the second type interface function and the corresponding first type interface function are the same;
compiling all the second type interface functions based on the API provided by the second framework to generate a packaging library;
recompiling the source code of the artificial intelligence application based on the packaging library and the API provided by the second framework so as to generate the artificial intelligence application running on the second framework.
2. The processing method of artificial intelligence application based on open source deep learning framework according to claim 1, characterized in that the processing method further comprises the steps of:
acquiring all data definitions in the first framework used in source codes of the artificial intelligence application;
packaging all the data definitions into the packaging library.
3. The method for processing artificial intelligence application based on open source deep learning framework according to claim 1 or 2,
the first framework is Caffe;
the first type of interface function comprises at least one of the following functions:
net- > Forward, Net- > copyTrainedLayerFrom, Net- > reset, Net- > input _ blob () and Net- > output _ blob ().
4. The processing method for artificial intelligence application based on open source deep learning framework according to claim 3, characterized in that the second framework is Tengine or Ncnn.
5. The processing method of artificial intelligence application based on open source deep learning framework according to claim 2, characterized in that the processing method further comprises the steps of:
acquiring basic data used in the source code of the artificial intelligence application; the basic data comprises a header file name, a data name, the name of the first type of interface function, the parameter type of the first type of interface function and the number of the first type of interface function;
packaging all the basic data into the packaging library.
6. A processing system of artificial intelligence application based on an open source deep learning framework relates to two open source deep learning frameworks which are a first framework and a second framework respectively, and the artificial intelligence application is developed based on the first framework;
characterized in that the processing system comprises:
an obtaining module, configured to obtain all interface functions used in the source code of the artificial intelligence application in the first framework, where the interface functions are first-class interface functions;
the realization module is used for realizing the function of each first-class interface function again based on the API provided by the second framework so as to obtain a second-class interface function corresponding to each first-class interface function; the interfaces of the second type interface function and the corresponding first type interface function are the same;
the packaging module is used for compiling all the second type interface functions based on the API provided by the second framework so as to generate a packaging library;
and the compiling module is used for recompiling the source code of the artificial intelligence application based on the packaging library and the API provided by the second framework so as to generate the artificial intelligence application running on the second framework.
7. The processing system for artificial intelligence applications based on an open source deep learning framework of claim 6,
the acquisition module is further used for acquiring all data definitions in the first framework used in the source code of the artificial intelligence application;
the packaging module is also used for packaging all the data definitions into the packaging library.
8. The processing system for artificial intelligence application based on open source deep learning framework according to claim 6 or 7,
the first framework is Caffe;
the first type of interface function comprises at least one of the following functions:
net- > Forward, Net- > copyTrainedLayerFrom, Net- > reset, Net- > input _ blob () and Net- > output _ blob ().
9. The processing system for artificial intelligence application based on open source deep learning framework of claim 8, characterized in that said second framework is Tengine or Ncnn.
10. The processing system for artificial intelligence applications based on an open source deep learning framework of claim 7,
the acquisition module is also used for acquiring basic data used in the source code of the artificial intelligence application; the basic data comprises a header file name, a data name, the name of the first type of interface function, the parameter type of the first type of interface function and the number of the first type of interface function;
the packaging module is also used for packaging all the basic data into the packaging library.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911182343.3A CN110989995A (en) | 2019-11-27 | 2019-11-27 | Processing method and system for artificial intelligence application based on open source deep learning framework |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911182343.3A CN110989995A (en) | 2019-11-27 | 2019-11-27 | Processing method and system for artificial intelligence application based on open source deep learning framework |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110989995A true CN110989995A (en) | 2020-04-10 |
Family
ID=70087205
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911182343.3A Pending CN110989995A (en) | 2019-11-27 | 2019-11-27 | Processing method and system for artificial intelligence application based on open source deep learning framework |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110989995A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113010181A (en) * | 2021-03-24 | 2021-06-22 | 北京百度网讯科技有限公司 | Deployment method and device of operators in deep learning framework and electronic equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107179916A (en) * | 2016-03-09 | 2017-09-19 | 腾讯科技(深圳)有限公司 | A kind of application development method, device, equipment and development frame |
CN108154238A (en) * | 2017-12-25 | 2018-06-12 | 东软集团股份有限公司 | Moving method, device, storage medium and the electronic equipment of machine learning flow |
CN108279881A (en) * | 2018-02-11 | 2018-07-13 | 深圳竹信科技有限公司 | Cross-platform realization framework based on deep learning predicted portions and method |
US20180330277A1 (en) * | 2017-05-10 | 2018-11-15 | Petuum Inc. | System and Methods for Distributed Machine Learning with Multiple Data Sources, Multiple Programming Languages or Frameworks, and Multiple Devices or Infrastructures |
CN110163345A (en) * | 2019-05-09 | 2019-08-23 | 腾讯科技(深圳)有限公司 | A kind of Processing with Neural Network method, apparatus, equipment and medium |
-
2019
- 2019-11-27 CN CN201911182343.3A patent/CN110989995A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107179916A (en) * | 2016-03-09 | 2017-09-19 | 腾讯科技(深圳)有限公司 | A kind of application development method, device, equipment and development frame |
US20180330277A1 (en) * | 2017-05-10 | 2018-11-15 | Petuum Inc. | System and Methods for Distributed Machine Learning with Multiple Data Sources, Multiple Programming Languages or Frameworks, and Multiple Devices or Infrastructures |
CN108154238A (en) * | 2017-12-25 | 2018-06-12 | 东软集团股份有限公司 | Moving method, device, storage medium and the electronic equipment of machine learning flow |
CN108279881A (en) * | 2018-02-11 | 2018-07-13 | 深圳竹信科技有限公司 | Cross-platform realization framework based on deep learning predicted portions and method |
CN110163345A (en) * | 2019-05-09 | 2019-08-23 | 腾讯科技(深圳)有限公司 | A kind of Processing with Neural Network method, apparatus, equipment and medium |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113010181A (en) * | 2021-03-24 | 2021-06-22 | 北京百度网讯科技有限公司 | Deployment method and device of operators in deep learning framework and electronic equipment |
US11531529B2 (en) | 2021-03-24 | 2022-12-20 | Beijing Baidu Netcom Science Technology Co., Ltd. | Method and electronic device for deploying operator in deep learning framework |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108885580B (en) | Combination of mode-driven responses in real-time data stream programming | |
CN110399133B (en) | JavaScript code optimization method based on front-end byte code technology | |
CN104298496B (en) | data analysis type software development framework system | |
US11314808B2 (en) | Hybrid flows containing a continous flow | |
CN104932905A (en) | Automatic code generation method from AADL to C language | |
CN112199086A (en) | Automatic programming control system, method, device, electronic device and storage medium | |
CN113031932B (en) | Project development method, apparatus, electronic device and storage medium | |
CN109976735A (en) | One kind being based on the visual knowledge mapping algorithm application platform of web | |
CN102779044B (en) | The dissection process system of expression formula and analyzing and processing method | |
CN103150200A (en) | Computer language transformation system and transformation method from C language to MSVL (Modeling, Simulation and Verification Language) | |
CN110941655A (en) | Data format conversion method and device | |
CN110109658A (en) | A kind of ROS code generator and code generating method based on formalized model | |
CN107015840B (en) | Quick execution method and quick execution system of resolver | |
CN114064601A (en) | Storage process conversion method, device, equipment and storage medium | |
CN103793653B (en) | A kind of program dependence based on tree optimization analyzes method and system | |
Sanchez et al. | Bigraphical modelling of architectural patterns | |
CN102929853A (en) | DCS (Data Communication System) project data generating system and method based on Excel sheet association | |
CN110989995A (en) | Processing method and system for artificial intelligence application based on open source deep learning framework | |
CN117009038B (en) | Graph computing platform based on cloud native technology | |
US20090328001A1 (en) | Unloadable managed code | |
CN106681781A (en) | Implementation method and system for real-time computing service | |
CN116756338A (en) | AR assembly guidance-oriented process knowledge graph construction method and system | |
CN115758789A (en) | Software architecture design and architecture transmission method of complex real-time embedded system | |
CN115469860A (en) | Method and system for automatically generating demand-to-software field model based on instruction set | |
CN102999323B (en) | A kind ofly generate the method for object code, the method for data processing and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200410 |