CN112508750A

CN112508750A - Artificial intelligence teaching device, method, equipment and storage medium

Info

Publication number: CN112508750A
Application number: CN202110143929.XA
Authority: CN
Inventors: 刘廷瑞; 邵秋实; 王汝佳
Original assignee: Beijing United Weishi Technology Co ltd
Current assignee: Beijing United Weishi Technology Co ltd
Priority date: 2021-02-03
Filing date: 2021-02-03
Publication date: 2021-03-16

Abstract

The application discloses an artificial intelligence teaching device, method, equipment and storage medium, and relates to the technical field of artificial intelligence teaching. The artificial intelligence teaching device includes: the system comprises an artificial intelligence learning module, a model control module, a resource scheduling module and a display processing module; the artificial intelligence learning module comprises at least one of the following sub-modules: a digital recognition learning submodule, a voice recognition learning submodule, a face recognition learning submodule, a two-dimensional (2D) human body posture estimation learning submodule and a three-dimensional (3D) human body posture estimation learning submodule; and each sub-module in the artificial intelligence learning module is used for providing a processing function of a neural network model based on human-computer interaction. By utilizing the method and the device, the education and teaching work of artificial intelligence can be intuitively and conveniently carried out.

Description

Artificial intelligence teaching device, method, equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence education, and in particular, to an artificial intelligence teaching apparatus, method, device, and storage medium.

Background

With the integration and development of Artificial Intelligence (AI) technology and the traditional field, the AI technology has become one of the most interesting technologies in the world, and has influenced the development of the world. Because the artificial intelligence technology covers a wide range of professional knowledge, a plurality of application fields, an abstract concept and a complex and difficult processing process, it is generally considered that the learning of the artificial intelligence technology is mainly concentrated in enterprises and colleges and research institutes with professional interfaces, the learning aims at mastering the application method of the artificial intelligence, for example, how to construct and optimize the architecture of the neural network, how to obtain training samples at low cost, how to optimize the training method of the neural network, and the like, and finally, a needed prediction model is generated and used in industrial application. The above learning process is a kind of conventional learning.

In practice, the artificial intelligence technology has been continuously permeated into the daily life of people, and the artificial intelligence technology appears in various new business states and new modes, so that the artificial intelligence technology is convenient for people to work, live, go out and the like. Under the background, how to popularize and popularize knowledge such as concepts and principles of artificial intelligence and the like enables people to intuitively understand the development and change of the environment, and is a problem that educators need to think and solve.

However, most of the existing artificial intelligence education and teaching platforms only pay attention to input and output of applications, do not pay attention to knowledge popularization of the artificial intelligence principle, and cannot allow learners to learn knowledge behind the applications. For example, for students of low ages who are in the knowledge enlightenment stage, the artificial intelligence concept can not be logically understood by only looking at the input and output, and the aim of education enlightenment is difficult to achieve in practice.

Disclosure of Invention

An artificial intelligence teaching apparatus, method, device and storage medium are provided.

According to a first aspect of the present application, there is provided an artificial intelligence teaching device comprising: the system comprises an artificial intelligence learning module, a model control module, a resource scheduling module and a display processing module; wherein the content of the first and second substances,

the artificial intelligence learning module comprises at least one of the following sub-modules: a digital recognition learning submodule, a voice recognition learning submodule, a face recognition learning submodule, a two-dimensional (2D) human body posture estimation learning submodule and a three-dimensional (3D) human body posture estimation learning submodule; each sub-module in the artificial intelligence learning module is used for providing a processing function of a neural network model based on human-computer interaction, and the processing function of the neural network model based on human-computer interaction comprises a processing process and a processing result of the neural network model according to a file input by a user;

the model control module is used for performing model reasoning on the file input by the user through a corresponding neural network model to obtain a prediction result and providing the prediction result to the artificial intelligence learning module;

the resource scheduling module is used for scheduling computing resources for the artificial intelligence learning module and the model control module;

and the display processing module is used for displaying the processing process and the processing result provided by the artificial intelligence learning module on a display screen.

According to the artificial intelligence teaching device of the embodiment of the application, optionally, the processing procedure corresponding to each sub-module in the artificial intelligence learning module includes: a file of user input displayed on the display screen and related content of the neural network model displayed on the display screen; the processing result corresponding to each sub-module in the artificial intelligence learning module comprises: and displaying the prediction result of the neural network model on a display screen.

According to the artificial intelligence teaching device of the embodiment of the application, optionally, the processing procedure corresponding to each sub-module in the artificial intelligence learning module further includes: and displaying the text description and/or the image display of the processing function corresponding to the sub-module on the display screen.

Optionally, the artificial intelligence teaching device according to the embodiment of the application, wherein the user input file corresponding to the digital recognition learning submodule includes at least one of: pre-storing a local image containing numbers, an image containing numbers shot by a camera device, and an image containing numbers input by a hand-drawing device; the relevant content of the neural network model corresponding to the digital recognition learning submodule comprises: the name and/or use of the first neural network model corresponding to the number recognition learning submodule; the prediction result of the neural network model corresponding to the number recognition learning submodule comprises: and the first neural network model carries out digital recognition processing on the file input by the user and outputs the processed file.

Optionally, the file input by the user corresponding to the speech recognition learning submodule includes a pre-stored local audio file and/or an audio file collected through a radio device; the relevant content of the neural network model corresponding to the speech recognition learning submodule comprises: the name and/or use of a second neural network model corresponding to the speech recognition learning submodule; the prediction result of the neural network model corresponding to the speech recognition learning submodule comprises: and the second neural network model carries out speech recognition processing on the file input by the user and then outputs the character.

Optionally, the artificial intelligence teaching device according to the embodiment of the application, wherein the file input by the user corresponding to the 2D human body posture estimation learning submodule includes at least one of: pre-storing a local image containing a person and an image containing the person shot by an image pickup device; the relevant content of the neural network model corresponding to the 2D human body posture estimation learning submodule comprises: the name and/or use of a third neural network model corresponding to the 2D human pose estimation learning submodule; the prediction result of the neural network model corresponding to the 2D human body posture estimation learning submodule comprises: and the third neural network model outputs two-dimensional position information of a plurality of human body posture key points after performing 2D human body posture estimation processing on the file input by the user.

Optionally, the artificial intelligence teaching device according to the embodiment of the application, wherein the file input by the user corresponding to the 3D human body posture estimation learning submodule includes at least one of: pre-storing a local image containing a person and an image containing the person shot by an image pickup device; the relevant content of the neural network model corresponding to the 3D human body posture estimation learning submodule comprises: the name and/or use of a fourth neural network model corresponding to the 3D human body pose estimation learning submodule; the prediction result of the neural network model corresponding to the 3D human body posture estimation learning submodule comprises: and the fourth neural network model is used for outputting the spatial position information of the plurality of human body joint points after performing 3D human body posture estimation processing on the file input by the user.

Optionally, the artificial intelligence teaching device according to the embodiment of the application, wherein the face recognition learning submodule includes: the system comprises a face detection learning unit, a face identification warehousing learning unit and a face retrieval learning unit; the file input by the user corresponding to the face detection learning unit comprises at least one of the following items: the method comprises the steps of pre-storing a local picture containing a human face, pre-storing a local video containing the human face, shooting the picture containing the human face through a camera device, and shooting the video containing the human face through the camera device; the relevant content of the neural network model corresponding to the face detection learning unit comprises: the name and/or use of a fifth neural network model corresponding to the face detection learning unit; the prediction result of the neural network model corresponding to the face detection learning unit comprises: the fifth neural network model is used for carrying out face feature matching processing on the file input by the user and outputting an image, and a face image is identified in the output image by a surrounding frame and confidence; the face identification warehousing learning unit is used for receiving a name label input by a user for the face image and storing the face image and the corresponding name label; the face retrieval learning unit is used for determining a target face according to a user instruction, retrieving faces similar to the target face from a plurality of face images stored by the face identification storage learning unit, and displaying the faces in a reverse order according to the similarity.

The artificial intelligence teaching device according to the embodiment of the application optionally further comprises an algorithm management module, which is used for accelerating the model reasoning process of the model control module through a reasoning engine.

According to the artificial intelligence teaching device of the embodiment of the application, optionally, the device further comprises a container management module, which is used for storing the artificial intelligence development framework and the neural network model adopted by the device.

According to a second aspect of the present application, there is provided an artificial intelligence teaching method based on the artificial intelligence teaching apparatus as described above, the artificial intelligence teaching method comprising:

receiving a selection instruction of a user through an artificial intelligence learning module, and receiving a to-be-processed file input by the user after the user selects at least one sub-module;

performing model reasoning on the file to be processed through a neural network model corresponding to the selected sub-module to obtain a prediction result, and providing the prediction result to the artificial intelligence learning module;

and displaying the processing process and the processing result corresponding to the selected sub-module on a display screen.

According to a third aspect of the present application, there is provided an electronic device comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein the content of the first and second substances,

the memory stores the artificial intelligence teaching device and instructions executable by the at least one processor to enable the at least one processor to perform the method as described above.

According to a fourth aspect of the present application, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing a computer to perform the method as described above.

The artificial intelligence teaching device can form logic-rich understanding to input, processing process and output of an artificial intelligence technology in common life scenes such as digital recognition, voice recognition, face recognition, 2D human posture estimation or 3D human posture estimation, improves the learning interest of students, and achieves the effect of AI knowledge education enlightenment in a simpler operation mode.

It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present application, nor do they limit the scope of the present application. Other features of the present application will become apparent from the following description.

Drawings

The drawings are included to provide a better understanding of the present solution and are not intended to limit the present application. Wherein:

FIG. 1 is a block diagram of an artificial intelligence teaching device according to an embodiment of the present application;

FIG. 2 is a block diagram of an artificial intelligence learning module according to an embodiment of the present application;

FIG. 3 is a block diagram of the operational flow of an artificial intelligence teaching method of an embodiment of the present application;

FIG. 4 is a diagram of the logical architecture of an artificial intelligence educational platform of an embodiment of the present application;

FIG. 5 is a schematic view of an operation flow of the industrial intelligence teaching platform according to the embodiment of the present application;

6-10 are schematic diagrams illustrating the processing procedure and the effect of the processing result of the number recognition learning submodule according to the embodiment of the present application;

11-15 are schematic diagrams illustrating the processing procedure and the effect of the processing result of the 2D human body posture estimation learning submodule according to the embodiment of the present application;

FIGS. 16-23 are schematic diagrams illustrating the processing procedure and the processing result of the speech recognition learning sub-module according to the embodiment of the present application;

FIG. 24 is a block diagram of an electronic device for implementing the artificial intelligence teaching method of the embodiments of the present application.

Detailed Description

The following description of the exemplary embodiments of the present application, taken in conjunction with the accompanying drawings, includes various details of the embodiments of the application for the understanding of the same, which are to be considered exemplary only. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the described embodiments can be made without departing from the scope and spirit of the application. Descriptions of well-known functions, constructions, and the like may be omitted from the following description for clarity and conciseness.

Fig. 1 shows a block diagram of an artificial intelligence teaching device provided in an embodiment of the present application, where the artificial intelligence teaching device includes: an artificial intelligence learning module 110, a model control module 120, a resource scheduling module 130 and a display processing module 140; wherein the artificial intelligence learning module 110 comprises one or more learning sub-modules.

Referring to fig. 2, the artificial intelligence learning module 110 may include at least one of the following learning sub-modules:

a digital recognition learning submodule 111, a voice recognition learning submodule 112, a face recognition learning submodule 113, a two-dimensional 2D human body posture estimation learning submodule 114 and a three-dimensional 3D human body posture estimation learning submodule 115;

moreover, each sub-module in the artificial intelligence learning module 110 is configured to provide a processing function of a neural network model based on human-computer interaction, where the processing function of the neural network model based on human-computer interaction includes a processing procedure and a processing result of the neural network model provided according to a file input by a user;

the model control module 120 is configured to perform model inference on the file input by the user through a corresponding neural network model to obtain a prediction result, and provide the prediction result to the artificial intelligence learning module;

the resource scheduling module 130 is configured to schedule computing resources for the artificial intelligence learning module 110 and the model control module 120;

the display processing module 140 is used for displaying the processing procedure and the processing result provided by the artificial intelligence learning module 110 on a display screen.

The artificial intelligence teaching device provided by the embodiment of the application can be carried on a computer (such as a server or a client) and used as a teaching principle platform, wherein an artificial intelligence learning module can provide processing functions of various artificial intelligence models, can be used for processing files (such as images, videos or other multimedia files and the like) selected by a user (such as a student) in a pre-stored file library or images or videos shot on site as processing objects, can be used for processing target objects in different application scenes according to requirements, such as identifying digital human faces, 2D human body postures or 3D human body postures in the images and outputting processing results, and can also be used for carrying out related knowledge contents (such as name purposes of the neural network models sampled in different scenes) of the neural network model involved in the processing process, and concept explanations, concepts and the like involved in the processing process, Graphic illustration, etc.) are presented on a display screen, which is equivalent to bringing different types of scene experiments into a classroom, so that students can not only visually observe and feel diversified processing capacity of the artificial intelligence technology in a learning field, but also can know specific knowledge points behind the concept of the artificial intelligence technology at a fine granularity through contents such as text explanation, legend display, etc. presented in the processing process.

Particularly, the artificial intelligence teaching device of the embodiment of the application can break through the original principle black box, and for students in an AI knowledge enlightenment stage, the artificial intelligence technology can form logical understanding on the input, processing and output in common life scenes such as digital recognition, voice recognition, face recognition, 2D human posture estimation or 3D human posture estimation, so that the learning interest of the students is promoted, and the AI knowledge enlightenment effect is achieved in a simpler operation mode.

Correspondingly, an embodiment of the present application further provides an artificial intelligence teaching method, which is based on the artificial intelligence teaching apparatus as described above, and referring to fig. 3, shows a flow chart of the artificial intelligence teaching method, and includes the following steps:

s101, receiving a selection instruction of a user through an artificial intelligence learning module, and receiving a to-be-processed file input by the user after the user selects at least one sub-module;

s102, performing model reasoning on the file to be processed through a neural network model corresponding to the selected sub-module to obtain a prediction result, and providing the prediction result to the artificial intelligence learning module;

s103, displaying the processing process and the processing result corresponding to the selected sub-module on a display screen.

By using the artificial intelligence teaching method of the embodiment of the application, a user such as a student can select interested sub-modules from the sub-modules suitable for various application scenes and autonomously select multimedia files such as pictures or videos to be processed, if a camera device is installed, images can be shot on site to serve as processing objects, visual experience of learners is enhanced, and because the system can display artificial intelligence knowledge related in the processing process on a screen, the system not only can visually observe and feel diversified processing capacity of the artificial intelligence technology in a learning site, but also can know knowledge points related to the artificial intelligence technology in different application scenes in a fine-grained manner, and learn the artificial intelligence knowledge more visually, neatly and meticulously.

In some embodiments of the present application, optionally, the processing procedure corresponding to each sub-module in the artificial intelligence learning module includes: a file of user input displayed on the display screen and related content of the neural network model displayed on the display screen; the processing result corresponding to each sub-module in the artificial intelligence learning module comprises: and displaying the prediction result of the neural network model on a display screen.

For example, a document to be processed, such as an image, determined by a user, such as a student, may be displayed on the screen, and also, related contents of a neural network model used to process the image (such as a name, a use, a model structure, a training manner, annotation data, and the like of the neural network) may be displayed on the screen, and a processing result, such as recognized numerals or characters, may be displayed on the screen. The advantage of this process is that the visual display can make students visually see the processing objects, the processing process and the processing results, which is beneficial to make students combine knowledge and application closely.

In some embodiments of the present application, optionally, the processing procedure corresponding to each sub-module in the artificial intelligence learning module further includes: and displaying the text description and/or the image display of the processing function corresponding to the sub-module on the display screen. For example, for the speech recognition learning sub-module, the "speech sound wave diagram" can be visually presented by pictures, and the word interpretation processing procedure can be used, for example:

"Speech analysis will convert each frame waveform into multi-dimensional vector (sound wave picture displayed on screen)"

"the acoustic model analyzes each audio frame to obtain the corresponding status code (status code of sound wave chart displayed on screen)"

"analyzing the status code to obtain the corresponding phoneme or pinyin (displaying pinyin on screen)"

"analyzing phonemes to obtain words (words displayed on screen)"

"to group words to compose sentences (sentences displayed on the screen)".

The advantage of handling like this is, can improve artificial intelligence teaching's visual degree to a great extent, and the student can directly perceivedly perceive the processing content and the implementation process of artificial intelligence technique, greatly improves the interest level of study exploration.

In some embodiments of the present application, optionally, the file of the user input corresponding to the number recognition learning sub-module includes at least one of: pre-storing a local image containing numbers, an image containing numbers shot by a camera device, and an image containing numbers input by a hand-drawing device; the relevant content of the neural network model corresponding to the digital recognition learning submodule comprises: the name and/or use of the first neural network model corresponding to the number recognition learning submodule; the prediction result of the neural network model corresponding to the number recognition learning submodule comprises: and the first neural network model carries out digital recognition processing on the file input by the user and outputs the processed file.

In some embodiments of the present application, optionally, the file corresponding to the speech recognition learning sub-module and input by the user includes an audio file pre-stored locally and/or an audio file collected by a radio device; the relevant content of the neural network model corresponding to the speech recognition learning submodule comprises: the name and/or use of a second neural network model corresponding to the speech recognition learning submodule; the prediction result of the neural network model corresponding to the speech recognition learning submodule comprises: and the second neural network model carries out speech recognition processing on the file input by the user and then outputs the character.

In some embodiments of the present application, optionally, the file of the user input corresponding to the 2D human posture estimation learning submodule includes at least one of: pre-storing a local image containing a person and an image containing the person shot by an image pickup device; the relevant content of the neural network model corresponding to the 2D human body posture estimation learning submodule comprises: the name and/or use of a third neural network model corresponding to the 2D human pose estimation learning submodule; the prediction result of the neural network model corresponding to the 2D human body posture estimation learning submodule comprises: and the third neural network model outputs two-dimensional position information of a plurality of human body posture key points after performing 2D human body posture estimation processing on the file input by the user.

In some embodiments of the present application, optionally, the file of the user input corresponding to the 3D human posture estimation learning submodule includes at least one of: pre-storing a local image containing a person and an image containing the person shot by an image pickup device; the relevant content of the neural network model corresponding to the 3D human body posture estimation learning submodule comprises: the name and/or use of a fourth neural network model corresponding to the 3D human body pose estimation learning submodule; the prediction result of the neural network model corresponding to the 3D human body posture estimation learning submodule comprises: and the fourth neural network model is used for outputting the spatial position information of the plurality of human body joint points after performing 3D human body posture estimation processing on the file input by the user.

In some embodiments of the present application, optionally, the face recognition learning sub-module includes: the system comprises a face detection learning unit, a face identification warehousing learning unit and a face retrieval learning unit; the file input by the user corresponding to the face detection learning unit comprises at least one of the following items: the method comprises the steps of pre-storing a local picture containing a human face, pre-storing a local video containing the human face, shooting the picture containing the human face through a camera device, and shooting the video containing the human face through the camera device; the relevant content of the neural network model corresponding to the face detection learning unit comprises: the name and/or use of a fifth neural network model corresponding to the face detection learning unit; the prediction result of the neural network model corresponding to the face detection learning unit comprises: the fifth neural network model is used for carrying out face feature matching processing on the file input by the user and outputting an image, and a face image is identified in the output image by a surrounding frame and confidence; the face identification warehousing learning unit is used for receiving a name label input by a user for the face image and storing the face image and the corresponding name label; the face retrieval learning unit is used for determining a target face according to a user instruction, retrieving faces similar to the target face from a plurality of face images stored by the face identification storage learning unit, and displaying the faces in a reverse order according to the similarity.

The first to fifth neural network models mentioned in the above embodiments may be suitable trained neural network models.

The input, the adopted neural network model and the output of a plurality of sub-modules are provided through a plurality of embodiments, the multi-module intelligent simulation experiment system can be used as scene experiments of different applications, the interactivity is enhanced, the practical ability of students is improved, different types of scene experiments are brought into a classroom, the students can be aware of the specific application of artificial intelligence in different industries and lives, the visual experience of the artificial intelligence principle is brought to the students, and the exploration of the knowledge of the artificial intelligence principle by the students is deepened.

In some embodiments of the present application, optionally, the artificial intelligence teaching device further includes: and the algorithm management module is used for accelerating the model reasoning process of the model control module through a reasoning engine and shortening the response time of the system.

In some embodiments of the present application, optionally, the artificial intelligence teaching device further includes: and the container management module is used for storing the artificial intelligence development framework and the neural network model adopted by the device.

The artificial intelligence teaching device of this application embodiment can satisfy the requirement of installing the deployment fast, can set up artificial intelligence education and teaching platform fast conveniently, and education institutions such as greatly reduced school insert artificial intelligence education and teaching platform's technical threshold.

Various implementations and operational procedures of embodiments of the present application are described in detail below with specific examples.

Fig. 4 is a diagram schematically illustrating a logical architecture of an artificial intelligence education platform according to an embodiment of the present application, the platform including: the system comprises an artificial intelligence learning module, a model control module, a resource scheduling module, a display processing module, an algorithm management module and a container management module, wherein the display processing module is not shown.

Specifically, the artificial intelligence learning module comprises the following 5 sub-modules: the system comprises a digital recognition learning submodule, a voice recognition learning submodule, a face recognition learning submodule, a 2D human body posture estimation learning submodule and a 3D human body posture estimation learning submodule.

The model control module includes the following sub-modules: the system comprises a digital recognition control submodule, a voice recognition control submodule, a face recognition control submodule, a 2D human posture estimation control submodule, a 3D human posture estimation control submodule, a comprehensive algorithm processing submodule, an image segmentation processing submodule, a voice recognition processing submodule, a semantic recognition processing submodule, a face recognition processing submodule, an image synthesis processing submodule and a feature recognition processing submodule.

The algorithm management module can comprise an algorithm container layer and an inference engine OpenVINO;

the container management module can comprise a plurality of neural network models and an artificial intelligence open source framework, wherein the neural network models comprise a Recurrent Neural Network (RNN), a Convolutional Neural Network (CNN), a long-short term memory artificial neural network (LSTM), a generative countermeasure network (GAN) and the like; open source frames such as Tensorflow, Pyorch, etc.

The resource scheduling module may include: device virtualization layer, distributed data system DaaS, X86 server, network, legacy storage, network attached storage NAS, and minicomputers.

The artificial intelligence education platform architecture adopts the concept of top layer design, can flexibly match artificial intelligence education teaching use scenes of different scales, and forms a solution of integrating calculation force support, algorithm, model and application. The platform middle layer is provided with a model control module, an algorithm management module and a container management module. The model control module comprises various models required by a service layer and can be gradually expanded along with the development of services; the algorithm management module adopts an OpenVINO inference engine acceleration model to infer and containerize common algorithms; the container management module is built into a common open source framework such as Tensorflow, Pythrch, etc. The underlying resource scheduling module may use system resources more efficiently.

As an example, fig. 5 schematically shows a flow chart of the usage of the artificial intelligence education platform of the embodiment of fig. 4, a user can log in through an account password, and then select a sub-module of interest from 5 learning sub-modules provided by the artificial intelligence learning module, and start learning and experiencing with the visual content provided by the sub-module. Logging may be exited after use.

The following description will be given taking an example in which the student selects the "number recognition learning submodule". Fig. 6 to 10 are schematic diagrams showing the processing procedure and the effect of the processing result provided by the number recognition learning sub-module, in which the contents of the screen display include an upper "illustration portion" and a lower "caption portion".

Further, the illustrated portion includes "original resources" on the left and "reasoning results" on the right, where the boxes of the original resources may display the content of the entered file, such as images selected to be pre-stored locally (containing the number "1") for uploading to the platform system as shown in fig. 7 and 8, and images written and uploaded in-situ by students through a tablet (containing the number "3") as shown in fig. 9.

Further, the corresponding word description part of the number recognition learning submodule can include the following contents:

inputting a number

Including numbers in the image, or by hand-writing numbers on a drawing board

② selecting a pre-training model

The LeNet convolution neural network model is proposed by Yan LeCun, the father of CNN, the model is mainly used for hand-written characters Identification and classification of

And identifying the number.

The inference results of the number recognition learning submodule, 1 and 3 respectively, are shown in the right box, as in fig. 8 and 10.

As yet another example, the following is described by taking as an example that a student selects a "2D human pose estimation learning submodule". 11-15 schematically illustrate the processing and effect of the processing results provided by the 2D human pose estimation learning sub-module. Fig. 11, 12, and 13 show a case where a pre-stored video file is selected for processing (1 person is shown in fig. 12, and a plurality of persons are shown in fig. 13), and fig. 14 and 15 show a case where a camera is used to capture an image in the field for processing.

The text description of the 2D body pose estimation learning sub-module may include two parts, the first part is an explanation of the related concept and processing function, purpose, etc. of the "2D body pose estimation", and the second part is related content of the neural network model used, for example, as follows:

this experiment mainly demonstrated a 2D human body posture using an optimized MobileNet V1 network for feature extraction Estimating a network

The pose of each person in the image is detected. The human trunk comprises key points and connecting lines between the key points

The pose contains 18 points: ear, eye, nose, neck, shoulder, eyebrow, elbow, wrist, hip, knee, ankle

Input video

② image preprocessing

Dividing the video into frames, wherein the image of each frame is required to meet the requirement of a training model

③ reasoning

Reasoning is carried out based on a model of a Mobilenet network and an ssd target detection algorithm

Fourthly, returning the result

Returning to 18 key points of the human body.

It can be seen that in fig. 13, 14 and 15, the right prediction result marks a plurality of human key points in the form of "points" and "lines", the left original resource and the right prediction result are displayed simultaneously, and the students can understand the processing of the model more easily through the intuitive comparison; and the related content of the model processing process is simultaneously given below the display screen, so that students can conveniently know the knowledge behind the module.

As still another example, the following description takes an example in which a student selects a "speech recognition learning submodule". Fig. 16-23 schematically illustrate the processing procedure and the effect of the processing result provided by the speech recognition learning sub-module. Fig. 16 and 17 show a display screen for selecting a local pre-stored audio file, and in other embodiments, a section of voice can be selectively recorded for field recognition; fig. 18-23 show a process demonstration of multiple processing steps for recognizing audio as text, using rich, intuitive, and visual speech spectrograms and text annotations, some of which are listed below:

speech analysis converts each frame waveform into a multi-dimensional vector (displaying a phonogram, e.g., FIGS. 18 and 19)

The acoustic model analyzes each audio frame to obtain a corresponding state code (the shape of the sound wave diagram is displayed beside the sound wave diagram) State code, as shown in FIG. 20)

Analyzing the status code to obtain the corresponding phoneme or pinyin (displaying pinyin beside the phonogram, as shown in FIG. 21)

Analyzing the phoneme to obtain words (displaying words beside the sound wave chart, as shown in FIG. 22)

The words are composed into sentences (sentences are displayed beside the sound wave picture, as shown in figure 23).

In addition, a text description part of the speech recognition learning submodule is also displayed at the same time, and the text description part can comprise the following contents:

recording voice

② speech spectrogram

Frame transition state code

State code conversion phonetic alphabet

Fifthly, phonetic conversion words

Sixthly, converting the words into sentences.

It can be seen that the process of processing the speech recognition task by the acoustic model is illustrated pictorially in fig. 18-23 by the sound wave diagrams and the text explanation of the different stages, and the processing by the model can be more easily understood and more impressive by the students through the intuitive comparison.

As still another example, the following description takes an example in which a student selects a "face recognition learning submodule". Specifically, after the 'face recognition learning submodule' is selected, three learning units can appear, which are respectively: the system comprises a face detection learning unit, a face identification warehousing learning unit and a face retrieval learning unit, wherein the function, the processing process and the processing result of the learning unit can be respectively displayed on a page corresponding to each learning unit, and the displayable text content is listed as follows:

A. face recognition learning submodule

a) Face detection learning unit

Input image

Preprocessing an input image into a size meeting the requirements of a model, wherein the size comprises conversion, scaling, center cutting and regularization;

② selecting a pre-training model

By using neural network models built by other persons or teams who solve the same problem, we no longer need to train from scratch The method can solve the same problem by randomly establishing a model, and can solve the problem by using the model as a starting point Solving other problems;

③ extraction of characteristics

According to the weight trained in advance, the convolutional neural network extracts image features and then classifies the image features to match human face features;

fourthly, returning the result

Returning the confidence coefficient and the frame information of the face of the input image, and finding out the face in the image;

b) face identification warehousing learning unit

Input image

② face detection

Completing boundary detection, and returning the detected face picture [ size 48 x 48 ] in the image]Acquiring a face image;

③ markingLabel (Bao)

Labeling the face photos, and marking the names corresponding to the face photos;

fourthly, finished warehousing

Clicking to put in storage, and finishing putting in storage of the face photos extracted from the images;

c) face retrieval learning unit

Input image

② face detection

Detecting a face in an image to be retrieved and finding out key points;

comparing with photos in photo library

Comparing the key points of the face detected in the last step with the key points of the face marked in the face library, and cosine-calculating Similarity;

fourthly, displaying the result

And selecting the first 5 faces according to the similarity reverse order.

It can be seen that three learning units (a face detection learning unit, a face identification warehousing learning unit and a face retrieval learning unit) in the face recognition learning submodule are mutually associated in processing function and processing logic, a face in an image is detected by the face detection learning unit, name labels of all faces are recorded by the face identification warehousing learning unit, and finally, when the input image to be detected contains the warehoused face, the face retrieval learning unit can output the face with high similarity and can also output the corresponding name, so that the face recognition function is realized. The face recognition learning submodule displays the realization principle and the process of the face recognition function step by step and step by step through the three learning units, and students can learn and experience the knowledge related to the artificial intelligence technology in the man-machine interaction process.

In other embodiments of the present application, learning sub-modules of other categories or functions may also be provided, so as to broaden the learner's cognition in the application field of the artificial intelligence technology, for example: the character recognition learning submodule, the mask recognition learning submodule, the automatic garbage bottle sorting learning submodule, the pest recognition learning submodule and the like. These are merely exemplary learning sub-modules, and the types of modules that can be added in practical applications are not limited thereto.

As a different example, the following exemplarily gives the content of the corresponding "text description" of some learning sub-modules, and the student is displayed under the screen when operating, so as to learn and understand the processing process of the neural network model on the input file.

B. Character recognition learning submodule

Input image

Uploading pictures containing English words

② preprocessing the picture

Scaling an image to meet requirements of a pre-processing model

Third, text detection

Adopting mask-RCNN character detection model, inputting metadata information (height, width), outputting character frame and confidence Thermodynamic diagram and text with fixed size of frames of degree, classification and all classifications

Coding

Using the result of the last step as input to generate coded text

Decoding

Decoding the encoded text to generate text characters

Sixthly, returning the result.

C. Mask recognition learning submodule

Input image

② image preprocessing

Processing the input image by scaling or the like to a size required for the face detection model

Third, detecting human face

Extracting the face features by adopting a Mobilenet neural network and returning the face features

Detecting mask

Recognizing the mask according to the mask detection model based on the face detected in the last step

And fifthly, outputting the result.

D. Automatic sorting learning submodule of garbage bottle

Input image

② preprocessing the picture

Scaling pictures to the size required by the model

③ loading garbage bottle neural network model

The model is obtained based on a ssd _ initiation _ v2 pre-training model and 200 marked pictures training, and is basically Can correctly identify the color of the plastic bottle, including transparent color, green color, white color, blue color, orange color and unknown color

Fourthly, reasoning

Inputting the preprocessed image into the network model and starting reasoning

And fifthly, returning the result.

E. Disease and pest identification learning submodule

Input image

② preprocessing the picture

Processing input pictures into formats required by pest neural network model

Loading pest and disease neural network model

The model is trained by adopting a target detection algorithm based on a deep learning network and mainly used for finishing rice diseases Automatic identification of disease states including bacterial leaf blight, brown spot and black smut

Fourthly, reasoning

And fifthly, returning the result.

F.3D human posture estimation learning submodule

The 2D human body posture estimation is to show the two-dimensional coordinates of human body joint points

The 3D human body posture estimation is to estimate the three-dimensional coordinates of human body joint points from pictures or videos

Application of 3D human body posture estimation technology to game, animation, behavior capture system, behavior understanding and other scenes

Input image

② image preprocessing

Reading the image frame by frame and inputting the current frame into the next step model

③ reasoning

Loading a 3D human body posture estimation model, inputting 256 x 448 images of the previous step, and then carrying out reasoning

Fourthly, outputting the result

And displaying the three-dimensional data of the human body in the image into a 2D window.

In other embodiments of the application, the contents such as the videos, the images, the characters and the like can be set in a targeted manner according to the situation, so that the actual application requirements are met. For students of different ages, the operation and the display of different modules are beneficial to obtaining the enlightenment effect of artificial intelligence principle knowledge of different degrees.

According to an embodiment of the present application, an electronic device and a readable storage medium are also provided. Fig. 24 is a block diagram of an electronic device according to an artificial intelligence teaching method according to an embodiment of the present application. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present application that are described and/or claimed herein.

As shown in fig. 24, the electronic apparatus includes: one or more processors 1001, memory 1002, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components are interconnected using different buses and may be mounted on a common motherboard or in other manners as desired. The processor may process instructions for execution within the electronic device, including instructions stored in or on the memory to display Graphical information for a Graphical User Interface (GUI) on an external input/output device, such as a display device coupled to the Interface. In other embodiments, multiple processors and/or multiple buses may be used, along with multiple memories and multiple memories, as desired. Also, multiple electronic devices may be connected, with each device providing portions of the necessary operations (e.g., as a server array, a group of blade servers, or a multi-processor system).

The memory 1002 is a non-transitory computer readable storage medium provided herein. The memory stores instructions executable by the at least one processor to cause the at least one processor to perform the artificial intelligence teaching method provided herein. The non-transitory computer readable storage medium of the present application stores computer instructions for causing a computer to perform the artificial intelligence teaching method provided by the present application.

The memory 1002, which is a non-transitory computer readable storage medium, may be used to store non-transitory software programs, non-transitory computer executable programs, and modules, such as program instructions/modules (e.g., modules or sub-modules shown in fig. 1 or 2) corresponding to the artificial intelligence teaching method in the embodiments of the present application. The processor 1001 executes various functional applications of the server and data processing by executing non-transitory software programs, instructions, and modules stored in the memory 1002, that is, implements the artificial intelligence teaching method in the above method embodiments.

The memory 1002 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function; the storage data area may store data created from analysis of the search result processing use of the electronic device, and the like. Further, the memory 1002 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid state storage device. In some embodiments, the memory 1002 may optionally include memory located remotely from the processor 1001, which may be connected to the analysis processing electronics of the search results over a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The electronic device corresponding to the artificial intelligence teaching method in the embodiment of the application may further include: an input device 1003 and an output device 1004. The processor 1001, the memory 1002, the input device 1003 and the output device 1004 may be connected by a bus or other means, and the embodiment of fig. 24 in this application is exemplified by the bus connection. The input device 1003 may receive input numeric or character information and generate key signal inputs related to user settings and function control of the electronic device for analysis processing of search results, such as an input device like a touch screen, a keypad, a mouse, a track pad, a touch pad, a pointer, one or more mouse buttons, a track ball, a joystick, etc. The output devices 1004 may include a display device, auxiliary lighting devices (e.g., LEDs), and tactile feedback devices (e.g., vibrating motors), among others. The Display device may include, but is not limited to, a Liquid Crystal Display (LCD), a Light Emitting Diode (LED) Display, and a plasma Display. In some implementations, the display device can be a touch screen.

Various implementations of the systems and techniques described here can be realized in digital electronic circuitry, Integrated circuitry, Application Specific Integrated Circuits (ASICs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.

These computer programs (also known as programs, software applications, or code) include machine instructions for a programmable processor, and may be implemented using high-level procedural and/or object-oriented programming languages, and/or assembly/machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or device (e.g., magnetic discs, optical disks, memory, Programmable Logic Devices (PLDs)) used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (Cathode Ray Tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the internet.

The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present application may be executed in parallel, sequentially, or in different orders, as long as the desired results of the technical solutions disclosed in the present application can be achieved, and the present invention is not limited herein.

The above-described embodiments should not be construed as limiting the scope of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An artificial intelligence teaching device, comprising: the system comprises an artificial intelligence learning module, a model control module, a resource scheduling module and a display processing module; wherein the content of the first and second substances,

2. The artificial intelligence teaching apparatus of claim 1 wherein,

the processing process corresponding to each sub-module in the artificial intelligence learning module comprises the following steps: a file of user input displayed on the display screen and related content of the neural network model displayed on the display screen;

the processing result corresponding to each sub-module in the artificial intelligence learning module comprises: and displaying the prediction result of the neural network model on a display screen.

3. The artificial intelligence teaching apparatus of claim 2 wherein,

the processing process corresponding to each sub-module in the artificial intelligence learning module further comprises: and displaying the text description and/or the image display of the processing function corresponding to the sub-module on the display screen.

4. The artificial intelligence teaching apparatus of claim 2 or 3 wherein,

the user input file corresponding to the digital recognition learning submodule comprises at least one of the following items: pre-storing a local image containing numbers, an image containing numbers shot by a camera device, and an image containing numbers input by a hand-drawing device;

the relevant content of the neural network model corresponding to the digital recognition learning submodule comprises: the name and/or use of the first neural network model corresponding to the number recognition learning submodule;

the prediction result of the neural network model corresponding to the number recognition learning submodule comprises: and the first neural network model carries out digital recognition processing on the file input by the user and outputs the processed file.

5. The artificial intelligence teaching apparatus of claim 2 or 3 wherein,

the files input by the user corresponding to the voice recognition learning submodule comprise audio files prestored in the local and/or audio files collected through radio equipment;

the relevant content of the neural network model corresponding to the speech recognition learning submodule comprises: the name and/or use of a second neural network model corresponding to the speech recognition learning submodule;

the prediction result of the neural network model corresponding to the speech recognition learning submodule comprises: and the second neural network model carries out speech recognition processing on the file input by the user and then outputs the character.

6. The artificial intelligence teaching apparatus of claim 2 or 3 wherein,

the file input by the user corresponding to the 2D human body posture estimation learning submodule comprises at least one of the following items: pre-storing a local image containing a person and an image containing the person shot by an image pickup device;

the relevant content of the neural network model corresponding to the 2D human body posture estimation learning submodule comprises: the name and/or use of a third neural network model corresponding to the 2D human pose estimation learning submodule;

the prediction result of the neural network model corresponding to the 2D human body posture estimation learning submodule comprises: and the third neural network model outputs two-dimensional position information of a plurality of human body posture key points after performing 2D human body posture estimation processing on the file input by the user.

7. The artificial intelligence teaching apparatus of claim 2 or 3 wherein,

the file input by the user corresponding to the 3D human body posture estimation learning submodule comprises at least one of the following items: pre-storing a local image containing a person and an image containing the person shot by an image pickup device;

the relevant content of the neural network model corresponding to the 3D human body posture estimation learning submodule comprises: the name and/or use of a fourth neural network model corresponding to the 3D human body pose estimation learning submodule;

the prediction result of the neural network model corresponding to the 3D human body posture estimation learning submodule comprises: and the fourth neural network model is used for outputting the spatial position information of the plurality of human body joint points after performing 3D human body posture estimation processing on the file input by the user.

8. The artificial intelligence teaching apparatus of claim 2 or 3 wherein,

the face recognition learning submodule comprises: the system comprises a face detection learning unit, a face identification warehousing learning unit and a face retrieval learning unit;

the file input by the user corresponding to the face detection learning unit comprises at least one of the following items: the method comprises the steps of pre-storing a local picture containing a human face, pre-storing a local video containing the human face, shooting the picture containing the human face through a camera device, and shooting the video containing the human face through the camera device;

the relevant content of the neural network model corresponding to the face detection learning unit comprises: the name and/or use of a fifth neural network model corresponding to the face detection learning unit;

the prediction result of the neural network model corresponding to the face detection learning unit comprises: the fifth neural network model is used for carrying out face feature matching processing on the file input by the user and outputting an image, and a face image is identified in the output image by a surrounding frame and confidence;

the face identification warehousing learning unit is used for receiving a name label input by a user for the face image and storing the face image and the corresponding name label;

the face retrieval learning unit is used for determining a target face according to a user instruction, retrieving faces similar to the target face from a plurality of face images stored by the face identification storage learning unit, and displaying the faces in a reverse order according to the similarity.

9. The artificial intelligence teaching device of any one of claims 1 to 3 further comprising:

the algorithm management module is used for accelerating the model reasoning process of the model control module through a reasoning engine;

and/or the presence of a gas in the gas,

and the container management module is used for storing the artificial intelligence development framework and the neural network model adopted by the device.

10. An artificial intelligence teaching method based on the artificial intelligence teaching apparatus according to any one of claims 1 to 9,

the artificial intelligence teaching method comprises the following steps:

11. An electronic device, comprising:

at least one processor; and

the memory stores the artificial intelligence teaching device and instructions executable by the at least one processor to enable the at least one processor to perform the method of claim 10.

12. A non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the method of claim 10.