CN114529854A - TinyML-based children language early education implementation method and system - Google Patents
TinyML-based children language early education implementation method and system Download PDFInfo
- Publication number
- CN114529854A CN114529854A CN202210156950.8A CN202210156950A CN114529854A CN 114529854 A CN114529854 A CN 114529854A CN 202210156950 A CN202210156950 A CN 202210156950A CN 114529854 A CN114529854 A CN 114529854A
- Authority
- CN
- China
- Prior art keywords
- model
- furniture
- tinyml
- language
- lightweight
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 34
- 240000007651 Rubus glaucus Species 0.000 claims abstract description 28
- 235000011034 Rubus glaucus Nutrition 0.000 claims abstract description 28
- 235000009122 Rubus idaeus Nutrition 0.000 claims abstract description 28
- 238000001514 detection method Methods 0.000 claims abstract description 21
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000012549 training Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 230000000366 juvenile effect Effects 0.000 claims description 4
- 230000009471 action Effects 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 2
- 238000010801 machine learning Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 6
- 230000002452 interceptive effect Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000003278 mimic effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/29—Graphical models, e.g. Bayesian networks
- G06F18/295—Markov models or related models, e.g. semi-Markov models; Markov random fields; Networks embedding Markov models
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/20—Education
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Tourism & Hospitality (AREA)
- Educational Technology (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Electrically Operated Instructional Devices (AREA)
Abstract
The invention discloses a method and a system for realizing language early education of children based on TinyML, belonging to the technical field of machine learning, edge calculation and data processing, aiming at solving the technical problem of realizing language teaching of children under the condition of limited accompanying time of parents, and adopting the technical scheme that: the method is characterized in that in a household environment, an edge computing device raspberry group 4B is matched with a lightweight target detection model NanoDet-Plus, real-time input is carried out through a portable camera, the lightweight target detection model NanoDet-Plus identifies furniture in an input video stream, an identification result is subjected to explanatory description through a Markov model and an expert system, and the edge computing device raspberry group 4B plays an obtained literal explanation through a loudspeaker, so that the purpose of language teaching of children is achieved.
Description
Technical Field
The invention relates to the technical field of machine learning, edge calculation and data processing, in particular to a method and a system for realizing language early education of children based on TinyML.
Background
According to the research of children's language experts, 0-6 years old is the key period for human to learn language, and parents should pay attention to the education method at home during this period. Studies have shown that around the age of an infant, parents should begin conversational interaction with the child. Around 12-18 months, the caring adult needs to make a descriptive interpretation of an item, rather than merely repeat the name of an item individually. After 18 months, there is a big outbreak of children's vocabulary, and at this stage parents ask the children questions after describing a scene, and promote the children's knowledge of the scene's items.
For the learning of foreign languages, on one hand, the prime period of learning foreign languages by children is seized, on the other hand, an environment is created for the children to learn foreign languages, the requirement of an interactive environment cannot be met by pure video music, and the interactive learning environment is more important in language learning because the children cannot effectively establish connection between article names and words.
Due to the increased work pressure today, young parents lack the time and energy to interact with their children. Therefore, the technical problem to be solved urgently at present is how to realize language teaching of children under the condition of limited accompanying time at home.
Disclosure of Invention
The invention provides a children language early education implementation method and system based on TinyML, and aims to solve the problem of how to realize language teaching of children under the condition of limited accompanying time at home.
The technical task of the invention is realized in the following way, the method for realizing language early education for children based on TinyML is characterized in that in a household environment, edge computing equipment raspberry group 4B is matched with a lightweight target detection model NanoDet-Plus, real-time input is carried out through a portable camera worn by children, the lightweight target detection model NanoDet-Plus identifies furniture in an input video stream, explanatory description is carried out on an identification result through a Markov model and an expert system, and the edge computing equipment raspberry group 4B plays a literal explanation obtained through a loudspeaker, so that the purpose of language teaching for children is realized.
Preferably, the method is specifically as follows:
preprocessing the data of the furniture database: establishing a furniture database by using the existing furniture data pictures, and processing the furniture data pictures;
establishing a lightweight target detection model NanoDet-Plus: constructing a lightweight target detection model NanoDet-Plus by using deep separable convolution and combining improved FCOS and Generalized Focal Loss; in order to obtain a better training result, loading pre-trained model parameters;
training on a GPU device: training is carried out on GPU equipment to obtain an existing furniture identification model;
lightweight existing furniture identification models: quantizing the output model to obtain a lightweight model suitable for the edge equipment;
deploying the existing furniture identification model after being lightened to a raspberry pie 4B;
a markov model and an expert system are set on raspberry pi 4B.
Preferably, the weight of the lightweight existing furniture identification model is quantized into UINT8 type by FP32, and the range of the quantized weight is [0,255 ]; the existing furniture identification models quantified are:
xfloat＝xscale*xquantized；
if the weight range of the existing furniture identification model is [ -1,1]Then xscaleIs composed of
More preferably, the depth separable convolution refers to a separation of a convolution kernel of 3x3 into a combination of a depth separable convolution of 3x3 and a convolution of 1x 1.
Preferably, the markov model is composed of a state and an action, the state is composed of a set of objects, when the object outside the state is identified by the lightweight existing furniture identification model, a signal is input into the state, the original state is changed into a new state, in the secondary process, the interpretation of an expert system is called, and the position information description of each object is output, such as that of a television on the left side of a dining table.
Preferably, the language interpretation and report output by the expert system provides two reading modes of Chinese and English.
A children language early education system based on TinyML, which comprises,
the system comprises a first establishing module, a second establishing module and a third establishing module, wherein the first establishing module is used for establishing a furniture database by using the existing furniture data pictures and processing the furniture data pictures;
a second construction module, which is used for constructing a lightweight target detection model NanoDet-Plus by using deep separable convolution and combining improved FCOS and Generalized Focal local; in order to obtain a better training result, loading pre-trained model parameters;
the acquisition module is used for training on GPU equipment to acquire the existing furniture identification model;
the light weight module is used for quantizing the output model to obtain a light weight model suitable for the edge equipment;
the deployment module is used for deploying the lightweight existing furniture identification model to a raspberry pi 4B;
and the setting module is used for setting the Markov model and the expert system on the raspberry pi 4B.
Preferably, the working process of the system is as follows:
(1) transmitting real-time data by using a portable camera;
(2) the lightweight existing furniture recognition model deployed on the raspberry pi 4B is used for target detection;
(3) outputting the identified furniture name by the raspberry pi 4B;
(4) sequentially inputting the furniture into the Markov model according to the size of the identification frame;
(5) and describing the positions of the furniture by an expert system, and playing the output text description by Chinese or English.
An electronic device, comprising: a memory and at least one processor;
wherein the memory has stored thereon a computer program;
the at least one processor executes the memory-stored computer program such that the at least one processor performs the TinyML-based juvenile language early education implementation method as described above.
A computer-readable storage medium having stored thereon a computer program executable by a processor to implement the TinyML-based juvenile language early education implementation method as described above.
The original target recognition network can achieve a recognition result with higher precision only after being trained by large-scale computing equipment, but the situation is changed by the generation of the MobileNet, lightweight models such as the MobileNet and the like can achieve recognition precision similar to a large model by using a smaller model scale, a NanoDet-Plus model released in 2021 can achieve 34% of recognition precision in 80 types of recognition, and the operation speed of edge equipment reaches about 20 FPS.
The method and the system for realizing language early education of children based on TinyML have the following advantages:
under the condition of limited accompanying time at home, the language teaching device realizes the language teaching purpose for children;
the invention interacts with the attention of the children through the edge device, when the children stay in a certain scene for a period of time, the device can perform explanatory description on the scene, and can effectively create a bilingual environment for the children through the voice broadcast of the Chinese-English bilingual version and provide recognition models of a plurality of scenes, thereby realizing the purpose of multi-scene language teaching;
the method can identify various common furniture in a home environment, trains the existing network through GPU equipment to obtain a high-precision furniture identification model, deploys the trained model and parameters in a raspberry pie 4B + WM8960 audio module through a lightweight technology, and the raspberry pie has identification precision similar to that of the GPU equipment;
and fourthly, the invention realizes the application of TinyML in a complex scene by using an open-source common development language and a library, and provides a new method for language training of children.
Drawings
The invention is further described below with reference to the accompanying drawings.
FIG. 1 is a flow chart diagram of a children language early education implementation method based on TinyML;
fig. 2 is a flow chart of the working process of the children language early education system based on TinyML.
Detailed Description
The method and system for implementing early childhood language education based on TinyML according to the present invention are described in detail below with reference to the drawings and specific embodiments of the specification.
Example 1:
in the embodiment, the method for realizing language early education for children based on TinyML is implemented in a household environment, edge computing equipment raspberry group 4B is matched with a lightweight target detection model NanoDet-Plus, real-time input is carried out through a portable camera worn by children, the lightweight target detection model NanoDet-Plus identifies furniture in an input video stream, an identification result is interpretatively described through a Markov model and an expert system, and the edge computing equipment raspberry group 4B plays an obtained literary explanation through a loudspeaker to realize the purpose of language teaching for children; the method comprises the following specific steps:
s1, preprocessing furniture database data: establishing a furniture database by using the existing furniture data pictures, and processing the furniture data pictures;
s2, establishing a lightweight target detection model NanoDet-Plus: constructing a lightweight target detection model NanoDet-Plus by using deep separable convolution and combining improved FCOS and Generalized Focal Loss; in order to obtain a better training result, loading pre-trained model parameters;
s3, training on GPU equipment: training is carried out on GPU equipment to obtain an existing furniture identification model;
s4, lightening the existing furniture recognition model: quantizing the output model to obtain a lightweight model suitable for the edge equipment;
s5, deploying the existing furniture recognition model after being lightened to raspberry pi 4B;
s6, setting a Markov model and an expert system on the raspberry pi 4B.
In this embodiment, the weight of the light weight existing furniture recognition model in step S4 is quantized from FP32 to UINT8 type, and the range of the quantized weight is [0,255 ]; the existing furniture identification models quantified are:
xfloat＝xscale*xquantized；
if the weight range of the existing furniture identification model is [ -1,1]Then xscaleIs composed of
The depth separable convolution in step S2 of the present embodiment refers to separating a convolution kernel of 3x3 into a combination of a depth separable convolution of 3x3 and a convolution of 1x 1.
The markov model in step S6 of this embodiment is composed of a state and an action, where the state is composed of a set of objects, and when the lightweight existing furniture recognition model recognizes an object outside the state, a signal is input to the state, the original state is changed to a new state, and in the next process, the interpretation of the expert system is invoked, and a description of the location information of each object, such as the left side of a table with a television, is output.
The embodiment provides two reading modes of Chinese and English for the literal explanation and broadcasting output by the expert system.
Example 2:
the children language early education system based on TinyML in the embodiment comprises,
the system comprises a first establishing module, a second establishing module and a third establishing module, wherein the first establishing module is used for establishing a furniture database by using the existing furniture data pictures and processing the furniture data pictures;
a second construction module, which is used for constructing a lightweight target detection model NanoDet-Plus by using deep separable convolution and combining improved FCOS and Generalized Focal local; in order to obtain a better training result, loading pre-trained model parameters;
the acquisition module is used for training on GPU equipment to acquire the existing furniture identification model;
the light weight module is used for quantizing the output model to obtain a light weight model suitable for the edge equipment;
the deployment module is used for deploying the lightweight existing furniture identification model to a raspberry pi 4B;
and the setting module is used for setting the Markov model and the expert system on the raspberry pi 4B.
The working process of the system is as follows:
(1) transmitting real-time data by using a portable camera;
(2) the lightweight existing furniture recognition model deployed on the raspberry pi 4B is used for target detection;
(3) outputting the identified furniture name by the raspberry pi 4B;
(4) sequentially inputting the furniture into the Markov model according to the size of the identification frame;
(5) and describing the positions of the furniture by an expert system, and playing the output text description by Chinese or English.
The system may also mimic the process of parents educating children to recognize items: for each type of identifiable labels, such as 'dining table', 'chair', 'carpet', etc., after the equipment successfully identifies the furniture type, the equipment carries out explanatory description on the furniture type, the process is realized by using a Markov model in combination with an expert system, the model can detect the mutual position information of a plurality of identification targets, and the scene is explained according to different position information. For example, when a chair appears alone, the device may report "this is a chair, one has four legs"; but the identified chair is next to the table, it will report "this is a table with 4chairs next to the table. The early education machine can also set interpretations of different language versions according to the requirements of parents to create a bilingual teaching environment for children, for example, the interpretation version of the example English is as follows: "This is a chair, a chair has four legs; this is a table, therae are 4 channels bed the table ".
In the use, child need wear portable camera on the head through the elastic cord, and camera direction and child's sight direction keep unanimous, and child's stay exceeds the certain time in a scene, and the early education machine will be taught child scene in front, and the size of object in the scene is followed to the teaching order, the big or small order of target detection discernment frame promptly, because in same space, the volume is big or small attracts child's attention easily more.
Example 3:
the present embodiment also provides an electronic device, including: a memory and a processor;
wherein the memory stores computer execution instructions;
the processor executes the computer-executable instructions stored by the memory, causing the processor to perform a TinyML-based early childhood language education implementation of any of the embodiments of the present invention.
The processor may be a Central Processing Unit (CPU), but may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), array of off-the-shelf programmable gates (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. The processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory may be used to store computer programs and/or modules, and the processor may implement various functions of the electronic device by executing or executing the computer programs and/or modules stored in the memory and invoking data stored in the memory. The memory may mainly 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, and the like; the storage data area may store data created according to the use of the terminal, and the like. The memory may also include high speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a memory only card (SMC), a Secure Digital (SD) card, a flash memory card, at least one disk storage period, a flash memory device, or other volatile solid state memory device.
Example 4:
embodiments of the present invention further provide a computer-readable storage medium, in which a plurality of instructions are stored, and the instructions are loaded by a processor, so that the processor executes the TinyML-based early childhood language education implementation in any embodiment of the present invention. Specifically, a system or an apparatus equipped with a storage medium on which software program codes that realize the functions of any of the above-described embodiments are stored may be provided, and a computer (or a CPU or MPU) of the system or the apparatus is caused to read out and execute the program codes stored in the storage medium.
In this case, the program code itself read from the storage medium can realize the functions of any of the above-described embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.
Examples of the storage medium for supplying the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (e.g., CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RYM, DVD-RW, DVD + RW), a magnetic tape, a nonvolatile memory card, and a ROM. Alternatively, the program code may be downloaded from a server computer via a communications network.
Further, it should be clear that the functions of any one of the above-described embodiments may be implemented not only by executing the program code read out by the computer, but also by causing an operating system or the like operating on the computer to perform a part or all of the actual operations based on instructions of the program code.
Further, it is to be understood that the program code read out from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion unit connected to the computer, and then causes a CPU or the like mounted on the expansion board or the expansion unit to perform part or all of the actual operations based on instructions of the program code, thereby realizing the functions of any of the above-described embodiments.
Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A child language early education implementation method based on TinyML is characterized in that in a home environment, an edge computing device raspberry group 4B is matched with a lightweight target detection model NanoDet-Plus, real-time input is conducted through a portable camera, the lightweight target detection model NanoDet-Plus identifies furniture in an input video stream, explanatory description is conducted on an identification result through a Markov model and an expert system, and an edge computing device raspberry group 4B plays an obtained literary explanation through a loudspeaker, so that the purpose of child language teaching is achieved.
2. The method for realizing early childhood language education based on TinyML as claimed in claim 1, wherein the method is as follows:
preprocessing the data of the furniture database: establishing a furniture database by using the existing furniture data pictures, and processing the furniture data pictures;
establishing a lightweight target detection model NanoDet-Plus: constructing a lightweight target detection model NanoDet-Plus by using deep separable convolution and combining improved FCOS and Generalized Focal Loss; loading the pre-trained model parameters;
training on a GPU device: training is carried out on GPU equipment to obtain an existing furniture identification model;
the existing furniture recognition model of lightweight: quantizing the output model to obtain a lightweight model suitable for the edge equipment;
deploying the existing furniture identification model after being lightened to a raspberry pie 4B;
a markov model and an expert system are set on raspberry pi 4B.
3. The method for realizing early childhood language education based on TinyML according to claim 2, wherein the weight of the lightweight existing furniture recognition model is quantized into UINT8 type by FP32, and the range of the quantized weight is [0,255 ]; the existing furniture identification models quantified are:
xfloat＝xscale*xquantized；
if the weight range of the existing furniture identification model is [ -1,1]Then xscaleIs composed of
4. The TinyML-based early childhood language education implementation method of claim 2, wherein the depth separable convolution is a combination of separating a convolution kernel of 3x3 into a depth separable convolution of 3x3 and a convolution of 1x 1.
5. The method for implementing early childhood language education based on TinyML according to claim 2, wherein the markov model is composed of a state and an action, the state is composed of a set of objects, when the object out of the state is identified by the lightweight existing furniture identification model, a signal is input to the state, the original state is changed into a new state, and in the secondary process, the interpretation of an expert system is called, and the position information description of each object is output.
6. The TinyML-based implementation method for the early childhood language according to any one of claims 2-5, wherein the textual interpretation and broadcast output by the expert system provide two reading modes of Chinese and English.
7. A children language early education system based on TinyML is characterized in that the system comprises,
the system comprises a first establishing module, a second establishing module and a third establishing module, wherein the first establishing module is used for establishing a furniture database by using the existing furniture data pictures and processing the furniture data pictures;
a second construction module, which is used for constructing a lightweight target detection model NanoDet-Plus by using deep separable convolution and combining improved FCOS and Generalized Focal local; in order to obtain a better training result, loading pre-trained model parameters;
the acquisition module is used for training on GPU equipment to acquire the existing furniture identification model;
the light weight module is used for quantizing the output model to obtain a light weight model suitable for the edge equipment;
the deployment module is used for deploying the lightweight existing furniture identification model to a raspberry pi 4B;
and the setting module is used for setting the Markov model and the expert system on the raspberry pi 4B.
8. The TinyML-based juvenile language early education system as claimed in claim 7, wherein the system works as follows:
(1) transmitting real-time data by using a portable camera;
(2) the lightweight existing furniture recognition model deployed on the raspberry pi 4B is used for target detection;
(3) outputting the identified furniture name by the raspberry pi 4B;
(4) sequentially inputting the furniture into the Markov model according to the size of the identification frame;
(5) and describing the positions of the furniture by an expert system, and playing the output text description by Chinese or English.
9. An electronic device, comprising: a memory and at least one processor;
wherein the memory has stored thereon a computer program;
the at least one processor executes the memory-stored computer program to cause the at least one processor to perform the TinyML-based early childhood language education implementation of any of claims 1-6.
10. A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and the computer program is executable by a processor to implement the TinyML-based juvenile language early education implementation method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210156950.8A CN114529854B (en) | 2022-02-21 | TinyML-based juvenile language early education implementation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210156950.8A CN114529854B (en) | 2022-02-21 | TinyML-based juvenile language early education implementation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114529854A true CN114529854A (en) | 2022-05-24 |
| CN114529854B CN114529854B (en) | 2024-05-31 |
Family
ID=
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103366618A (en) * | 2013-07-18 | 2013-10-23 | 梁亚楠 | Scene device for Chinese learning training based on artificial intelligence and virtual reality |
| CN111243599A (en) * | 2020-01-13 | 2020-06-05 | 网易有道信息技术（北京）有限公司 | Speech recognition model construction method, device, medium and electronic equipment |
| CN111833878A (en) * | 2020-07-20 | 2020-10-27 | 中国人民武装警察部队工程大学 | Chinese voice interaction non-inductive control system and method based on raspberry Pi edge calculation |
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103366618A (en) * | 2013-07-18 | 2013-10-23 | 梁亚楠 | Scene device for Chinese learning training based on artificial intelligence and virtual reality |
| CN111243599A (en) * | 2020-01-13 | 2020-06-05 | 网易有道信息技术（北京）有限公司 | Speech recognition model construction method, device, medium and electronic equipment |
| CN111833878A (en) * | 2020-07-20 | 2020-10-27 | 中国人民武装警察部队工程大学 | Chinese voice interaction non-inductive control system and method based on raspberry Pi edge calculation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shen et al. | Hugginggpt: Solving ai tasks with chatgpt and its friends in hugging face | |
| Aafaq et al. | Video description: A survey of methods, datasets, and evaluation metrics | |
| US11409945B2 (en) | Natural language processing using context-specific word vectors | |
| Achlioptas et al. | Referit3d: Neural listeners for fine-grained 3d object identification in real-world scenes | |
| Ramanishka et al. | Multimodal video description | |
| Aafaq et al. | Spatio-temporal dynamics and semantic attribute enriched visual encoding for video captioning | |
| Poliak et al. | Hypothesis only baselines in natural language inference | |
| Li et al. | TGIF: A new dataset and benchmark on animated GIF description | |
| Bowman et al. | A large annotated corpus for learning natural language inference | |
| Yang et al. | Corpus-guided sentence generation of natural images | |
| CN112104919B (en) | Content title generation method, device, equipment and computer readable storage medium based on neural network | |
| Rashtchian et al. | Collecting image annotations using amazon’s mechanical turk | |
| US11501082B2 (en) | Sentence generation method, sentence generation apparatus, and smart device | |
| Lin et al. | Generating multi-sentence lingual descriptions of indoor scenes | |
| CN112955911A (en) | Digital image classification and annotation | |
| CN106202056B (en) | Chinese word segmentation scene library update method and system | |
| Kumar et al. | A survey of evolution of image captioning techniques | |
| Varshney et al. | Unsupervised natural language inference using phl triplet generation | |
| Turkerud et al. | Image captioning using deep learning: text augmentation by paraphrasing via backtranslation | |
| Kalouli et al. | Curing the SICK and other NLI maladies | |
| CN114529854A (en) | TinyML-based children language early education implementation method and system | |
| Himakunthala et al. | Let’s think frame by frame with VIP: A video infilling and prediction dataset for evaluating video chain-of-thought | |
| CN114529854B (en) | TinyML-based juvenile language early education implementation method and system | |
| CN116186246A (en) | Public opinion text abstract generation method and related device based on PEGASUS and entity chain | |
| Bowman et al. | The snli corpus |
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 | ||
| GR01 | Patent grant |