CN112416301A - Deep learning model development method and device and computer readable storage medium - Google Patents
Deep learning model development method and device and computer readable storage medium Download PDFInfo
- Publication number
- CN112416301A CN112416301A CN202011283390.XA CN202011283390A CN112416301A CN 112416301 A CN112416301 A CN 112416301A CN 202011283390 A CN202011283390 A CN 202011283390A CN 112416301 A CN112416301 A CN 112416301A
- Authority
- CN
- China
- Prior art keywords
- deep learning
- model
- option
- learning model
- options
- 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
- 238000013136 deep learning model Methods 0.000 title claims abstract description 165
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000011161 development Methods 0.000 title claims abstract description 45
- 238000003860 storage Methods 0.000 title claims abstract description 22
- 238000012360 testing method Methods 0.000 claims abstract description 84
- 238000012549 training Methods 0.000 claims abstract description 45
- 238000002372 labelling Methods 0.000 claims abstract description 35
- 230000018109 developmental process Effects 0.000 claims description 56
- 238000013135 deep learning Methods 0.000 claims description 23
- 230000001133 acceleration Effects 0.000 claims description 21
- 238000001514 detection method Methods 0.000 claims description 13
- 238000007781 pre-processing Methods 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 10
- 238000012545 processing Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 238000012986 modification Methods 0.000 description 6
- 230000004048 modification Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F8/00—Arrangements for software engineering
- G06F8/20—Software design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Electrically Operated Instructional Devices (AREA)
Abstract
The invention discloses a deep learning model development method and a device thereof, and a computer readable storage medium, wherein the method comprises the following steps: setting up a deep learning model according to user settings aiming at a pre-provided model building option; acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set; training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain a trained deep learning model; and testing the trained deep learning model based on the self-defined model test options and corresponding user settings. The invention solves the problem of higher threshold of deep learning model development, realizes the improvement of the deep learning model development efficiency and reduces the threshold of the deep learning model development.
Description
Technical Field
The present application relates to the field of deep learning technologies, and in particular, to a method and an apparatus for developing a deep learning model, and a computer-readable storage medium.
Background
The development and application process of the deep learning model relates to multiple development stages such as data labeling, model building, model training, model testing and model deployment, and each development stage needs to depend on different tools, technologies, professionals and the like. In addition, deep learning model developers need to have development experience of corresponding platforms in the development process, and need a certain deep learning algorithm foundation. Therefore, the development threshold of the deep learning model is relatively high.
Disclosure of Invention
By providing the deep learning model development method and device and the computer readable storage medium, the embodiment of the application solves the problem of higher threshold of deep learning model development, improves the efficiency of deep learning model development and reduces the threshold of deep learning model development.
The embodiment of the application provides a deep learning model development method, which comprises the following steps:
setting up a deep learning model according to user settings aiming at a pre-provided model building option;
acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set;
training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain a trained deep learning model;
testing the trained deep learning model based on a self-defined model test option and corresponding user setting;
the model building option, the data labeling option, the model training option and the model testing option are provided through a graphical interface, and user settings are received through the graphical interface.
In one embodiment, the model building options include a model type option and a model size option; wherein the model type at least comprises detection, classification, segmentation and regression; the model sizes include at least large, medium and small.
In one embodiment, the model building options further include a model combination option; the step of building the deep learning model according to the user setting aiming at the pre-provided model building options comprises the following steps:
setting up a preset number of deep learning submodels according to user settings aiming at the pre-provided model type options and model size options;
and combining the deep learning submodels according to the user setting aiming at the pre-provided model combination options to obtain the deep learning model.
In an embodiment, the data in the data set to be labeled is a picture; the data annotation option comprises a picture zoom option.
In one embodiment, the model training options include a batch size setting option, a graphics processor acceleration option, and a data pre-processing option.
In one embodiment, the model test options include a single model test option, a graphics processor acceleration option, and a global test option; the step of testing the trained deep learning model comprises:
and after testing all the deep learning submodels, integrally testing the deep learning submodels by using the corresponding test data sets.
In an embodiment, the method further comprises:
compiling the tested deep learning model according to user settings aiming at the pre-provided model compiling options; wherein the model compilation option is provided via a graphical interface.
In one embodiment, the model compilation options include a quantized data set selection option, a graphics processor acceleration option, a translation file type option.
The embodiment of the application also provides a device, which comprises a processor, a memory and a deep learning model development program stored on the memory and capable of running on the processor, wherein when the deep learning model development program is executed by the processor, the steps of the deep learning model development method are realized.
The embodiment of the application also provides a computer readable storage medium, wherein a deep learning model development program is stored on the computer readable storage medium, and when being executed by a processor, the deep learning model development program realizes the steps of the deep learning model development method.
The technical scheme of the deep learning model development method and device and the computer readable storage medium provided in the embodiment of the application has at least the following technical effects:
because the deep learning model is set up according to the user setting aiming at the pre-provided model building options; acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set; training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain a trained deep learning model; testing the trained deep learning model based on a self-defined model test option and corresponding user setting; the method comprises the steps of providing a model building option, a data marking option, a model training option and a model testing option through a graphical interface, and receiving a technical means set by a user through the graphical interface. Therefore, the problem that the threshold for developing the deep learning model is high is effectively solved, the efficiency for developing the deep learning model is improved, and the threshold for developing the deep learning model is reduced.
Drawings
FIG. 1 is a schematic structural diagram of an apparatus according to an embodiment of the present disclosure;
FIG. 2 is a schematic flowchart of a deep learning model development method according to a first embodiment of the present application;
FIG. 3 is a flowchart illustrating a deep learning model development method according to a second embodiment of the present invention;
fig. 4 is a flowchart illustrating a third embodiment of the deep learning model development method according to the present application.
Detailed Description
In order to solve the problem that the threshold for developing the deep learning model is high, the deep learning model is set and built according to the user setting aiming at the pre-provided model building options; acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set; training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain a trained deep learning model; testing the trained deep learning model based on a self-defined model test option and corresponding user setting; the method comprises the steps of providing a model building option, a data marking option, a model training option and a model testing option through a graphical interface, and receiving a technical scheme set by a user through the graphical interface. The efficiency of developing the deep learning model is improved, and the threshold of developing the deep learning model is reduced.
For a better understanding of the above technical solutions, exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
Referring to fig. 1, it is a schematic diagram of a hardware structure of an apparatus involved in various embodiments of the present application, where the apparatus may include: processor 101, memory 102, input module 103, output module 104, and the like. Those skilled in the art will appreciate that the hardware configuration of the apparatus shown in fig. 1 does not constitute a limitation of the apparatus, which may include more or less components than those shown, or some components in combination, or a different arrangement of components.
The various components of the device are described in detail below with reference to fig. 1:
the processor 101 is a control center of the apparatus, connects various parts of the entire apparatus, and performs various functions of the apparatus or processes data by running or executing a program stored in the memory 102 and calling up the data stored in the memory 102, thereby monitoring the entire apparatus.
The memory 102 may be used to store various programs of the device as well as various data. The memory 102 mainly includes a program storage area and a data storage area, wherein the program storage area at least stores programs required for deep learning model development; the storage data area may store various data of the device. Further, the memory 102 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid state storage device.
The input module 103 may be used to input various data and user settings required for deep learning model development. Wherein user settings may be entered via a graphical interface.
The output module 104 can be used to output the results of the deep learning model development as well as to present pre-provided model development options to the user. Wherein the model development options can be output through a graphical interface.
In the embodiment of the present application, the processor 101 may be configured to call the deep learning model development program stored in the memory 102, and perform the following operations:
setting up a deep learning model according to user settings aiming at a pre-provided model building option;
acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set;
training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain a trained deep learning model;
testing the trained deep learning model based on a self-defined model test option and corresponding user setting;
the model building option, the data labeling option, the model training option and the model testing option are provided through a graphical interface, and user settings are received through the graphical interface.
In one embodiment, the model building options include a model type option and a model size option; wherein the model type at least comprises detection, classification, segmentation and regression; the model sizes include at least large, medium and small.
In one embodiment, the model building options further include a model combination option; the processor 101 may be configured to invoke a deep learning model development program stored in the memory 102 and perform the following operations:
setting up a preset number of deep learning submodels according to user settings aiming at the pre-provided model type options and model size options;
and combining the deep learning submodels according to the user setting aiming at the pre-provided model combination options to obtain the deep learning model.
In an embodiment, the data in the data set to be labeled is a picture; the data annotation option comprises a picture zoom option.
In one embodiment, the model training options include a batch size setting option, a graphics processor acceleration option, and a data pre-processing option.
In one embodiment, the model test options include a single model test option, a graphics processor acceleration option, and a global test option; the processor 101 may be configured to invoke a deep learning model development program stored in the memory 102 and perform the following operations:
and after testing all the deep learning submodels, integrally testing the deep learning submodels by using the corresponding test data sets.
In one embodiment, the processor 101 may be configured to invoke a deep learning model development program stored in the memory 102 and perform the following operations:
compiling the tested deep learning model according to user settings aiming at the pre-provided model compiling options; wherein the model compilation option is provided via a graphical interface.
In one embodiment, the model compilation options include a quantized data set selection option, a graphics processor acceleration option, a translation file type option.
According to the technical scheme, the deep learning model is set up according to the user setting aiming at the pre-provided model building options; acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set; training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain a trained deep learning model; testing the trained deep learning model based on a self-defined model test option and corresponding user setting; the method comprises the steps of providing a model building option, a data marking option, a model training option and a model testing option through a graphical interface, and receiving a technical means set by a user through the graphical interface. Therefore, the problem that the threshold for developing the deep learning model is high is effectively solved, the efficiency for developing the deep learning model is improved, and the threshold for developing the deep learning model is reduced.
For better understanding of the above technical solutions, the following detailed descriptions will be provided in conjunction with the drawings and the detailed description of the embodiments.
Referring to fig. 2, in a first embodiment of the present application, a deep learning model development method specifically includes the following steps:
step S110, building a deep learning model according to the user setting aiming at the pre-provided model building options.
In this embodiment, the model building options include a model type option and a model size option; wherein the model type at least comprises detection, classification, segmentation and regression; the model sizes include at least large, medium and small. The user can select a corresponding model type from the pre-provided model type options according to the service type of the deep learning model to be built, for example, if the deep learning model to be built is used for detecting pedestrians in a picture, the model type can be selected as detection. Meanwhile, the user can select a corresponding model size from the pre-provided model size options according to the size of the hardware resource of the user equipment, for example, if the user equipment is a server, the model size can be selected to be large.
In an embodiment, the model building options further comprise a model combination option. The user can set the model types and the model sizes of a plurality of sub models first, and then set the combination mode of the plurality of sub models according to the preset model combination options, so that the combination model formed by combining the plurality of sub models can be obtained. The combination mode of the plurality of submodels may be a plurality of modes such as series, parallel or combination of series and parallel. The user can cascade a plurality of sub-models according to the service requirement.
After user setting aiming at the pre-provided model building options is obtained, a corresponding deep learning model can be built according to the needs of the user. And when the deep learning model is built, a data labeling label set by a user can be further acquired, so that preparation is carried out for data labeling of the next step. For example, if the built deep learning model is used for detecting pedestrians in a picture, the user may set the data tagging label of the deep learning model as "pedestrian".
And step S120, acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set.
In this embodiment, after the deep learning model is built, the corresponding data set to be labeled can be obtained based on the built deep learning model. The data set to be annotated may be a data set to be annotated provided in advance, or a data set to be annotated input by a user. And after the corresponding data set of the data set to be labeled is obtained, labeling the data set to be labeled according to the user setting aiming at the data labeling option provided in advance. The label for data labeling may be a data labeling label preset by a user. And if the data in the data set to be labeled is a picture, the data labeling option further comprises a picture zooming option, that is, a user can zoom the picture under the condition that the target to be labeled is smaller or larger in the picture.
And step S130, training the deep learning model by using the labeled data set according to user settings aiming at the pre-provided model training options to obtain the trained deep learning model.
In this embodiment, the model training options include a batch size setting option, a graphics processor acceleration option, and a data pre-processing option. The user can set the corresponding training batch size (batch size) in the pre-provided batch size setting options according to the actual needs; a user can select whether to start Graphic Processor (GPU) acceleration training or not from pre-provided graphic processor acceleration options according to the condition of own equipment hardware resources; the user can also select whether the data needs to be preprocessed or not from the data preprocessing options provided in advance according to the self requirement so as to increase the disturbance of the training data and improve the robustness of the model. Taking data preprocessing of the picture data as an example, the data preprocessing may include random flipping, random brightness, random saturation, channel conversion, and the like. After the built deep learning model, the corresponding labeled data set and user settings for the pre-provided model training options are obtained, the labeled data set is used for training the deep learning model according to the user settings for the pre-provided model training options, and the trained deep learning model is obtained. In addition, the user can also choose to continue training the trained deep learning model so as to facilitate the iterative optimization of the model.
And step S140, testing the trained deep learning model based on the self-defined model testing options and corresponding user settings.
In this embodiment, the model test options include a test data set selection option, a single model test option, a graphics processor acceleration option, an overall test option, and a test result processing option. The user can select the model test option in a self-defined way according to the actual requirement. For example, for a deep learning model consisting of one sub-model, the user may select only the test data set selection option, the single model test option, and the graphics processor acceleration option. For a deep learning model composed of a plurality of sub-models, a user can select a test data set selection option, a single model test option, a graphic processor acceleration option, an overall test option and a test result processing option. The test data set can be a test data set provided in advance or a test data set input by a user; the graphic processor acceleration option is used for providing graphic processor acceleration for a user, and the user can select whether to start a graphic processor acceleration test according to own equipment hardware resources; the single model test option user tests a single sub model; the integral test option is used for integrally testing a deep learning model consisting of a plurality of sub models; the test result processing option can be used for processing the test result when the test result of one sub-model is needed to be used as training data of another sub-model, for example, the test result can be a picture to be scratched. After the trained deep learning model, the customized model test options and the corresponding user settings are obtained, the trained deep learning model can be tested by using the test data set.
In addition, the model building option, the data labeling option, the model training option and the model testing option can be provided through a graphical interface, and user settings are received through the graphical interface.
The method has the advantages that the deep learning model is set and built according to the user setting aiming at the pre-provided model building options; acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set; training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain a trained deep learning model; testing the trained deep learning model based on a self-defined model test option and corresponding user setting; the method comprises the steps of providing a model building option, a data marking option, a model training option and a model testing option through a graphical interface, and receiving a technical scheme set by a user through the graphical interface. Therefore, the problem that the threshold for developing the deep learning model is high is effectively solved, the efficiency for developing the deep learning model is improved, and the threshold for developing the deep learning model is reduced.
Referring to fig. 3, in a second embodiment of the present application, a deep learning model development method specifically includes the following steps:
and step S211, building a preset number of deep learning submodels according to the user setting aiming at the pre-provided model type options and model size options.
In this embodiment, a deep learning model combined by a plurality of submodels is required to be built by a user, so that a preset number of deep learning submodels are required to be built according to user settings for a pre-provided model type option and a model size option, that is, a plurality of submodels are built according to the model type and the model size of each submodel selected by the user. In this case, the data label of each sub-model may be preset for the subsequent data labeling.
And step S212, combining the deep learning submodels according to the user setting aiming at the pre-provided model combination options to obtain a deep learning model.
In this embodiment, after a plurality of deep learning submodels are built, the deep learning submodels may be combined according to user settings for pre-provided model combination options, that is, the plurality of deep learning submodels are combined according to a combination mode selected by a user, so as to obtain a deep learning model required by the user.
And step S220, acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set.
And step S230, training the deep learning model by using the labeled data set according to user settings aiming at the pre-provided model training options to obtain the trained deep learning model.
And step S241, testing all deep learning submodels based on the self-defined model testing options and corresponding user settings, and then integrally testing the deep learning submodels by using corresponding testing data sets.
In this embodiment, it is assumed that the deep learning submodels built by the user are a detection submodel a and a classification submodel B, respectively, and the detection submodel a is connected with the classification submodel B in series. The data label of the detection sub-model A is 'pedestrian', and the data label of the classification sub-model B is 'with safety helmet' or 'without safety helmet'. Firstly, data annotation of an original data set is required according to user operation, namely pedestrian annotation is carried out on the original data set to generate a pedestrian detection data set; then, performing model training on the detection sub-model A by using a pedestrian detection data set according to the operation of a user; testing the trained detection submodel A by using a test data set according to the operation of a user, and meanwhile, matting the detected pedestrian; classifying and marking whether the extracted pedestrian picture wears the safety helmet or not according to the operation of the user to generate a safety helmet classification data set; performing model training on the classification sub-model B by using the safety helmet classification data set according to user operation; then, testing the trained classification submodel B by using a test data set according to the operation of a user until the two submodels are trained and tested; and then, the deep learning model consisting of the detection sub-model A and the classification sub-model B is integrally tested by using a test data set, so that the development of the deep learning model can be completed.
The method has the beneficial effect that the development steps of the deep learning model of the multi-model combination are refined on the basis of the first embodiment. Therefore, the problem that the threshold for developing the deep learning model is high is further effectively solved, the efficiency for developing the deep learning model is improved, and the threshold for developing the deep learning model is reduced.
Referring to fig. 4, in a third embodiment of the present application, a deep learning model development method specifically includes the following steps:
step S310, building a deep learning model according to the user setting aiming at the pre-provided model building options.
And step S320, acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set.
And step S330, training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain the trained deep learning model.
And step S340, testing the trained deep learning model based on the self-defined model testing options and corresponding user settings.
And step S350, compiling the tested deep learning model according to the user setting aiming at the model compiling options provided in advance.
In this embodiment, the model compilation options include a quantized data set selection option, a graphics processor acceleration option, and a translated file type option. The quantitative data set can be a pre-provided quantitative data set or a quantitative data set input by a user and is used for quantifying the deep learning model parameters, so that the parameter capacity of the model is effectively reduced, and the calculation force requirement required by the running of the model is reduced; the graphic processor acceleration option is used for providing graphic processor acceleration for a user, and the user can select whether to start graphic processor accelerated compiling or not according to own equipment hardware resources; the translate file type option is used to provide a user with a variety of different file type selections to compile the tested deep learning model into a file format that can be run on an associated hardware accelerator card. Wherein the model compilation option may be provided through a graphical interface. After the tested deep learning model and the user setting for the pre-provided model compiling option are obtained, the tested deep learning model can be compiled, namely quantized, and then the model file is translated.
The method has the advantage that the step of compiling the deep learning model is supplemented on the basis of the first embodiment. Therefore, the problem that the threshold for developing the deep learning model is high is further effectively solved, the efficiency for developing the deep learning model is improved, and the threshold for developing the deep learning model is reduced.
Based on the same inventive concept, an embodiment of the present application further provides an apparatus, where the apparatus includes a processor, a memory, and a deep learning model development program that is stored in the memory and can be run on the processor, and when the deep learning model development program is executed by the processor, the deep learning model development program implements each process of the above embodiment of the deep learning model development method, and can achieve the same technical effect, and is not described here again to avoid repetition.
Since the apparatus provided in the embodiments of the present application is an apparatus used for implementing the method in the embodiments of the present application, based on the method described in the embodiments of the present application, a person skilled in the art can understand the specific structure and the variation of the apparatus, and thus details are not described herein again. All devices used in the methods of the embodiments of the present application are within the scope of the present application.
Based on the same inventive concept, an embodiment of the present application further provides a computer-readable storage medium, where a deep learning model development program is stored on the computer-readable storage medium, and when executed by a processor, the deep learning model development program implements the processes of the above deep learning model development method, and can achieve the same technical effects, and is not described herein again to avoid repetition.
Since the computer-readable storage medium provided in the embodiments of the present application is a computer-readable storage medium used for implementing the method in the embodiments of the present application, based on the method described in the embodiments of the present application, those skilled in the art can understand the specific structure and modification of the computer-readable storage medium, and thus details are not described herein. Any computer-readable storage medium that can be used with the methods of the embodiments of the present application is intended to be within the scope of the present application.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The usage of the words first, second and third, etcetera do not indicate any ordering. These words may be interpreted as names.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A method for deep learning model development, the method comprising:
setting up a deep learning model according to user settings aiming at a pre-provided model building option;
acquiring a corresponding data set to be labeled based on the built deep learning model, and labeling the data set to be labeled according to user setting aiming at a data labeling option provided in advance to obtain a labeled data set;
training the deep learning model by using the labeled data set according to user settings aiming at pre-provided model training options to obtain a trained deep learning model;
testing the trained deep learning model based on a self-defined model test option and corresponding user setting;
the model building option, the data labeling option, the model training option and the model testing option are provided through a graphical interface, and user settings are received through the graphical interface.
2. The deep learning model development method of claim 1, wherein the model building options comprise a model type option, a model size option; wherein the model type at least comprises detection, classification, segmentation and regression; the model sizes include at least large, medium and small.
3. The deep learning model development method of claim 2, wherein the model building options further comprise a model combination option; the step of building the deep learning model according to the user setting aiming at the pre-provided model building options comprises the following steps:
setting up a preset number of deep learning submodels according to user settings aiming at the pre-provided model type options and model size options;
and combining the deep learning submodels according to the user setting aiming at the pre-provided model combination options to obtain the deep learning model.
4. The deep learning model development method of claim 1, wherein the data in the dataset to be labeled is a picture; the data annotation option comprises a picture zoom option.
5. The method of deep learning model development of claim 1, wherein the model training options include a batch size setting option, a graphics processor acceleration option, a data pre-processing option.
6. The deep learning model development method of claim 3, wherein the model test options include a single model test option, a graphics processor acceleration option, an overall test option; the step of testing the trained deep learning model comprises:
and after testing all the deep learning submodels, integrally testing the deep learning submodels by using the corresponding test data sets.
7. The method of deep learning model development according to claim 1, the method further comprising:
compiling the tested deep learning model according to user settings aiming at the pre-provided model compiling options; wherein the model compilation option is provided via a graphical interface.
8. The method of deep learning model development of claim 7, wherein the model compilation options include a quantized data set selection option, a graphics processor acceleration option, a translated file type option.
9. An apparatus comprising a processor, a memory, and a deep learning model development program stored on the memory and executable on the processor, the deep learning model development program when executed by the processor implementing the steps of the deep learning model development method of any one of claims 1 to 8.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a deep learning model development program, which when executed by a processor implements the steps of the deep learning model development method according to any one of claims 1 to 8.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011122380 | 2020-10-19 | ||
| CN2020111223808 | 2020-10-19 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112416301A true CN112416301A (en) | 2021-02-26 |
Family
ID=74831393
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011283390.XA Pending CN112416301A (en) | 2020-10-19 | 2020-11-16 | Deep learning model development method and device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112416301A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113052328A (en) * | 2021-04-02 | 2021-06-29 | 上海商汤科技开发有限公司 | Deep learning model production system, electronic device, and storage medium |
| CN113505895A (en) * | 2021-08-05 | 2021-10-15 | 上海高德威智能交通***有限公司 | Machine learning engine service system, model training method and configuration method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109447248A (en) * | 2018-11-09 | 2019-03-08 | 深圳增强现实技术有限公司 | Deep learning platform and method |
| CN110689136A (en) * | 2019-09-06 | 2020-01-14 | 广东浪潮大数据研究有限公司 | Deep learning model obtaining method, device, equipment and storage medium |
| CN110991649A (en) * | 2019-10-28 | 2020-04-10 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Deep learning model building method, device, equipment and storage medium |
| CN111310934A (en) * | 2020-02-14 | 2020-06-19 | 北京百度网讯科技有限公司 | Model generation method and device, electronic equipment and storage medium |
| CN111612132A (en) * | 2020-05-20 | 2020-09-01 | 广东电网有限责任公司 | Artificial intelligence algorithm development system, training method, device and medium |
-
2020
- 2020-11-16 CN CN202011283390.XA patent/CN112416301A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109447248A (en) * | 2018-11-09 | 2019-03-08 | 深圳增强现实技术有限公司 | Deep learning platform and method |
| CN110689136A (en) * | 2019-09-06 | 2020-01-14 | 广东浪潮大数据研究有限公司 | Deep learning model obtaining method, device, equipment and storage medium |
| CN110991649A (en) * | 2019-10-28 | 2020-04-10 | 中国电子产品可靠性与环境试验研究所（（工业和信息化部电子第五研究所）（中国赛宝实验室）） | Deep learning model building method, device, equipment and storage medium |
| CN111310934A (en) * | 2020-02-14 | 2020-06-19 | 北京百度网讯科技有限公司 | Model generation method and device, electronic equipment and storage medium |
| CN111612132A (en) * | 2020-05-20 | 2020-09-01 | 广东电网有限责任公司 | Artificial intelligence algorithm development system, training method, device and medium |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113052328A (en) * | 2021-04-02 | 2021-06-29 | 上海商汤科技开发有限公司 | Deep learning model production system, electronic device, and storage medium |
| WO2022205835A1 (en) * | 2021-04-02 | 2022-10-06 | 上海商汤科技开发有限公司 | Deep learning model production system, electronic device and storage medium |
| CN113505895A (en) * | 2021-08-05 | 2021-10-15 | 上海高德威智能交通***有限公司 | Machine learning engine service system, model training method and configuration method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7059318B2 (en) | Learning data generation method and system for classifier learning with regional characteristics | |
| CN110046706B (en) | Model generation method and device and server | |
| Cruz et al. | Using convolutional neural networks in robots with limited computational resources: detecting NAO robots while playing soccer | |
| CN107045503A (en) | The method and device that a kind of feature set is determined | |
| CN109829371B (en) | Face detection method and device | |
| CN112416301A (en) | Deep learning model development method and device and computer readable storage medium | |
| CN111931179A (en) | Cloud malicious program detection system and method based on deep learning | |
| CN110599453A (en) | Panel defect detection method and device based on image fusion and equipment terminal | |
| JP2019159836A (en) | Learning program, learning method and learning device | |
| CN113535399A (en) | NFV resource scheduling method, device and system | |
| CN111373406A (en) | Accelerated simulation setup procedure using a priori knowledge extraction for problem matching | |
| CN103870563B (en) | It is determined that the method and apparatus of the theme distribution of given text | |
| CN114048816B (en) | Method, device, equipment and storage medium for sampling data of graph neural network | |
| CN109800776A (en) | Material mask method, device, terminal and computer readable storage medium | |
| CN111783812A (en) | Method and device for identifying forbidden images and computer readable storage medium | |
| CN108520532B (en) | Method and device for identifying motion direction of object in video | |
| CN112533060B (en) | Video processing method and device | |
| CN110716778B (en) | Application compatibility testing method, device and system | |
| CN111539390A (en) | Small target image identification method, equipment and system based on Yolov3 | |
| CN109063561B (en) | Formula identification calculation method and device | |
| CN114816719B (en) | Training method and device of multi-task model | |
| CN112990461A (en) | Method and device for constructing neural network model, computer equipment and storage medium | |
| KR102413588B1 (en) | Object recognition model recommendation method, system and computer program according to training data | |
| CN109145991A (en) | Image group generation method, image group generating means and electronic equipment | |
| CN111459774A (en) | Method, device and equipment for acquiring flow of application program and storage medium |
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 |