CN113450776A - Data enhancement method and system for improving crying detection model effect of baby - Google Patents
Data enhancement method and system for improving crying detection model effect of baby Download PDFInfo
- Publication number
- CN113450776A CN113450776A CN202010215286.0A CN202010215286A CN113450776A CN 113450776 A CN113450776 A CN 113450776A CN 202010215286 A CN202010215286 A CN 202010215286A CN 113450776 A CN113450776 A CN 113450776A
- Authority
- CN
- China
- Prior art keywords
- samples
- sample
- data
- negative
- positive
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 206010011469 Crying Diseases 0.000 title claims abstract description 43
- 238000001514 detection method Methods 0.000 title claims abstract description 22
- 230000000694 effects Effects 0.000 title claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000013528 artificial neural network Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 7
- 238000004088 simulation Methods 0.000 claims description 2
- 230000000474 nursing effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000036651 mood Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013529 biological neural network Methods 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000000873 masking effect Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000007958 sleep Effects 0.000 description 1
- 230000004382 visual function Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
- G10L25/30—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique using neural networks
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
- G10L2015/0635—Training updating or merging of old and new templates; Mean values; Weighting
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Signal Processing (AREA)
- Measuring And Recording Apparatus For Diagnosis (AREA)
Abstract
The invention provides a data enhancement method for improving the crying detection model effect of an infant, which comprises the following steps: s1, collecting sound samples as training data; s2, classifying the collected sound samples into positive samples and negative samples, wherein the positive samples are baby crying data in a quiet environment, and the negative samples are sound data except the baby crying; s3, adding a label corresponding to each type of voice data in the positive sample crying data and the negative sample voice data; and S4, obtaining new virtual samples and labels by linearly interpolating the samples and labels by using a hybrid enhancement method so as to increase the sample data. And enriching sample data and fitting a real scene by a mixed enhancement means.
Description
Technical Field
The invention relates to the technical field of audio processing, in particular to a data enhancement method and a data enhancement system for improving the crying detection model effect of a baby.
Background
Today's society is in the spotlight with regard to baby care, and the high cost of hiring caregivers such as nurses, etc. is more affordable for many average families. Some young parents who need to engage in nursing the infant are becoming more and more busy and less experienced in nursing the infant. When the infant care tool is handed to the old to nurse the infant, the problems that the infant cries and is not timely cared, the quilt is kicked when the infant sleeps, the body of the infant is not comfortable and cannot be timely known and the like can also occur due to the age, the body and the like of the old. Therefore, the intelligent infant nursing becomes a trend, and how to accurately and effectively detect the crying of the infant becomes a more and more concerned problem. In the prior art, a common processing method for audio sample data includes: 1, cropping, rotating, adding noise, changing pitch, etc. And 2, arbitrarily blocking or shielding the voice spectrogram.
However, the prior art has the following drawbacks:
1. although the sample size is increased by clipping, rotating, adding noise and the like, the change of original sample data is small, the increase of the richness of the sample is limited, and the sample has a larger difference with the sample of a real scene. Changing the pitch of the sound sample is not suitable for baby cry detection applications, because the baby is not in use of the language, and the expression of the baby's mood is largely reflected in the frequency of the sound emitted, so changing the pitch causes frequency changes that obscure the mood information in the sound segment, which is important in distinguishing between "talking" and crying "of the baby.
2. Occlusion or masking of the spectrogram can result in loss of critical pieces of information, such as crying short, incoherent samples.
Furthermore, the common terminology in the prior art is as follows:
1. machine vision: the computer simulates the visual function of human eyes through an optical device to capture the image of an objective object or scene, extracts key features from the image information through analysis, and then performs measurement or judgment, and is an important branch in the field of artificial intelligence development.
2. A neural network: the method simulates a mathematical model of the structure and the function of a biological neural network, obtains the capability of analyzing or expressing sample data by learning the internal rule of training the sample data, and can be applied to various application fields such as target detection, scene classification, character recognition and the like.
3. Deep learning: a process and method for training a neural network.
4. Overfitting: the deep neural network has stronger expression capability than the shallow neural network due to the complexity of the model, but has more parameters, so more training samples are required. If the scale of the training sample is small, the model can learn the rule of the training sample data too intensively, but the generalization capability is weak, the data reasoning result similar to the distribution of the training sample data is good, and the reasoning capability on other sample data is poor.
5. Data enhancement: the original training samples are processed to generate virtual samples different from the original training samples, so that the sample scale is expanded, the sample diversity is increased, the neural network generalization capability is improved, and the overfitting problem caused by the training samples is improved.
6. Voice spectrogram: and sound frequency domain information obtained through Fourier transform.
7. And (3) mixing and enhancing: the method is a method commonly used in the field of image processing, new samples and tag data are generated through linear interpolation, and the sample data scale is expanded.
Disclosure of Invention
In order to solve the above problems, the present invention is directed to:
1. the invention provides a hybrid enhancement idea applied to the field of machine vision and applied to baby cry detection, and the method expands the scale of a crying training sample without destroying key information of the crying sample and improves the over-fitting problem.
2. The collected baby crying training sample is closer to the real scene sample, and the generalization capability of the detection model is improved.
Specifically, the invention provides a data enhancement method for improving the effect of a detection model for baby crying, which comprises the following steps:
s1, collecting sound samples as training data;
s2, classifying the collected sound samples into positive samples and negative samples, wherein the positive samples are baby crying data in a quiet environment, and the negative samples are sound data except the baby crying;
s3, adding a label corresponding to each type of voice data in the positive sample crying data and the negative sample voice data;
and S4, obtaining new virtual samples and labels by linearly interpolating the samples and labels by using a hybrid enhancement method so as to increase the sample data.
In step S3, adding the labels corresponding to the tags respectively further includes: let x beiIndicating the superimposed sample data, label _ xiDenotes the corresponding specimen label, yjSample data representing the overlay, label _ yjRepresenting the corresponding label.
The step S4 further includes:
virtual sample g obtained by linear interpolationkAnd a corresponding label _ gkThe following were used:
gk＝γ*xi+(1-γ)*yj
γ＝rand(0.7，1.0)
wherein, sample xiAnd yjWith partial overlap, gamma is taken to be [0.7,1.0 ]]A random value in between.
A data enhancement system for improving the effect of a baby crying detection model comprises a collection unit, a classification unit and a data processing unit using a hybrid enhancement method, wherein the system is realized by adopting any one of the methods.
Thus, the present application has the advantages that: the hybrid enhancement means is used for the baby cry detection application, sample data are enriched through the hybrid enhancement means and are fit with real scenes, and the problems that the collection of cry samples capable of covering the scenes is very difficult due to various actual scenes and the labeling of samples collected under a complex environment is difficult are solved through a simple method and a simple system. Low cost and convenient upgrading and maintenance.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the principles of the invention.
FIG. 1 is a schematic flow diagram of the process of the present invention.
Fig. 2 is a schematic block diagram of the system of the present invention.
Detailed Description
In order that the technical contents and advantages of the present invention can be more clearly understood, the present invention will now be described in further detail with reference to the accompanying drawings.
As shown in fig. 1, the present invention relates to a data enhancement method for improving the effect of a crying detection model of an infant, the method comprises the following steps:
s1, collecting sound samples as training data;
s2, classifying the collected sound samples into positive samples and negative samples, wherein the positive samples are baby crying data in a quiet environment, and the negative samples are sound data except the baby crying;
s3, adding a label corresponding to each type of voice data in the positive sample crying data and the negative sample voice data;
and S4, obtaining new virtual samples and labels by linearly interpolating the samples and labels by using a hybrid enhancement method so as to increase the sample data.
In step S3, adding the labels corresponding to the tags respectively further includes: let x beiIndicating the superimposed sample data, label _ xiDenotes the corresponding specimen label, yjSample data representing the overlay, label _ yiRepresenting the corresponding label.
The step S4 further includes:
virtual sample g obtained by linear interpolationkAnd a corresponding label _ gkThe following were used:
gk＝γ*xi+(1-γ)*yj
γ＝rand(0.7，1.0)
wherein, sample xiAnd yiWith partial overlap, gamma is taken to be [0.7,1.0 ]]A random value in between.
The negative examples include at least one or more of the following: a sound sample other than the baby crying, a public environment sound sample, and silence.
The negative sample is preferably indoor and outdoor life and work environment data under normal conditions.
The method may further comprise:
and S5, sending the training data into a neural network training model, and training the voice data in the sample training set by using a deep neural network algorithm to obtain an acoustic detection model of the baby crying.
As shown in fig. 2, a data enhancement system for improving the effect of a baby cry detection model comprises a collecting unit for collecting sound samples as training data; the classification unit is used for classifying the collected sound samples into positive samples and negative samples, wherein the positive samples are baby crying data in a quiet environment, and the negative samples are sound data except the baby crying; the data processing unit is used for adding labels corresponding to the positive sample crying data and the negative sample sound data to each type of sound data in the positive sample crying data and the negative sample sound data respectively; and obtaining new virtual samples and labels by linear interpolation of the samples and the labels by using a hybrid enhancement method to increase sample data, wherein the system is realized by adopting any one of the methods.
Generally, the larger the sample data size of a training sample is, the better the learned model effect is, and the stronger the generalization ability is, however, in practical situations, many problems are faced to obtain enough sample data capable of covering all real scenes, and in more cases, a smaller-scale training sample data is used to learn a model, so how to obtain more high-quality sample data through a data enhancement method is very important.
The deep neural network-based baby cry detection model training needs to collect a large amount of baby cry data as positive sample data, but the baby is uncontrollable, and the practical scene that the baby cry occurs is various, so that the collection of the cry sample capable of covering the scenes is very difficult, and the sample collected under the complex environment is difficult to label, so that the positive sample collected by the method is the baby cry data under the quiet environment, the negative sample is the indoor and outdoor life and working environment data under the normal condition, and then the sample data is enriched by a mixed enhancement means and is attached to the real scene.
Hybrid enhancement is a method commonly used for image data enhancement, and it is a very effective means for infant crying sample enhancement to obtain new virtual samples and labels by linearly interpolating the samples and labels. Let x beiIndicating the superimposed sample data, label _ xiDenotes the corresponding specimen label, yjSample data representing the overlay, label _ yjVirtual samples g interpolated to represent corresponding labelskAnd a label _ gkThe following were used:
gk＝γ*xi+(1-γ)*yj
γ＝rand(0.7，1.0)
wherein, sample xiAnd yjWith partial overlap, gamma is taken to be [0.7,1.0 ]]The lower limit of 0.7 is an empirical value, and more key information pieces can be reserved.
Several scenarios can be simulated by hybrid enhancement:
(1) sample xiAnd yjAre all positive samples, sample gkMultiple crying sounds are simulated, and the diversity of positive samples is enriched;
(2) sample xiIs a positive sample, yjIs a negative sample, sample gkSimulation of occurrence of yjUnder the condition, the scene of the baby crying enriches the diversity of positive samples;
(3) sample xiAnd yjAre all negative samples, sample gkAnd simulating a complex negative sample scene.
Specifically, when positive and negative samples are superimposed as in (2) above, the superimposed sample yjSample data which is very close to the positive sample and difficult to classify, such as the baby dental language, is not included, otherwise, the false detection rate is high; (3) when negative samples are superimposed, if xiAnd yjThe classification capability of model learning on the samples can be enhanced and the false detection probability is reduced when the samples are difficult to classify; the interpolation method can also be used for superposing the samples for multiple times, and the superposition times are determined according to the quality of the superposed samples.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes may be made to the embodiment of the present invention by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A data enhancement method for improving the effect of a crying detection model of an infant, the method comprising the steps of:
s1, collecting sound samples as training data;
b2, classifying the collected sound samples into positive samples and negative samples, wherein the positive samples are baby crying data in a quiet environment, and the negative samples are sound data except the baby crying;
s3, adding a label corresponding to each type of voice data in the positive sample crying data and the negative sample voice data;
and S4, obtaining new virtual samples and labels by linearly interpolating the samples and labels by using a hybrid enhancement method so as to increase the sample data.
2. The method as claimed in claim 1, wherein the step S3 of adding labels respectively corresponding to the above mentioned data comprises the steps of: let x beiRepresenting superimposed samplesData, label _ xiDenotes the corresponding specimen label, yjSample data representing the overlay, label _ yjRepresenting the corresponding label.
3. The method as claimed in claim 2, wherein the step B4 further comprises:
virtual sample g obtained by linear interpolationkAnd a corresponding label _ gkThe following were used:
gk＝γ*xi+(1-γ)*yj
γ＝rand(0.7，1.0)
wherein, sample xiAnd yjWith partial overlap, gamma is taken to be [0.7,1.0 ]]A random value in between.
4. The method as claimed in claim 1, wherein the negative samples include at least one or more of the following types: a sound sample other than the baby crying, a public environment sound sample, and silence.
5. The method as claimed in claim 4, wherein the negative samples are selected from indoor and outdoor life and work environment data under normal conditions.
6. The method as claimed in claim 3, wherein the hybrid enhancement method simulates the following scenarios:
(1) sample xiAnd yjAre all positive samples, sample gkMultiple crying sounds are simulated, and the diversity of positive samples is enriched;
(2) sample xiIs a positive sample, yjIs a negative sample, sample gkSimulation of occurrence of yjUnder the condition, the scene of the baby crying enriches the diversity of positive samples;
(3) sample xiAnd yjAre all negative samples, sample gkAnd simulating a complex negative sample scene.
7. The method as claimed in claim 6, wherein the superimposed sample y is obtained by superimposing the positive and negative samples in the step (2)jSample data which is extremely close to the positive sample and difficult to classify is not included, so that high false detection rate is avoided; when (3) the negative samples are superposed, if xiAnd yjIf the negative samples are difficult to classify, the classification capability of the model learning on the samples is enhanced, so that the false detection probability is reduced; the linear interpolation method can also be used for superposing samples for multiple times, and the superposition times are determined according to the quality of the superposed samples.
8. The method as claimed in claim 7, wherein the sample data of the difficult classification includes dental language of the infant.
9. The method as claimed in claim 1, wherein the method further comprises:
and S5, sending the training data into a neural network training model, and training the voice data in the sample training set by using a deep neural network algorithm to obtain an acoustic detection model of the baby crying.
10. A data enhancement system for improving the effect of a baby crying detection model is characterized by comprising a collecting unit, a data processing unit and a data processing unit, wherein the collecting unit is used for collecting sound samples as training data; the classification unit is used for classifying the collected sound samples into positive samples and negative samples, wherein the positive samples are baby crying data in a quiet environment, and the negative samples are sound data except the baby crying; and a data processing unit using a hybrid enhancement method for adding a label corresponding to each type of voice data in the positive sample crying data and the negative sample voice data; and adding sample data by linearly interpolating the samples and labels using a hybrid enhancement method to obtain new virtual samples and labels, said system using the method of any of claims 1-9 above.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010215286.0A CN113450776A (en) | 2020-03-24 | 2020-03-24 | Data enhancement method and system for improving crying detection model effect of baby |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010215286.0A CN113450776A (en) | 2020-03-24 | 2020-03-24 | Data enhancement method and system for improving crying detection model effect of baby |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113450776A true CN113450776A (en) | 2021-09-28 |
Family
ID=77807479
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010215286.0A Pending CN113450776A (en) | 2020-03-24 | 2020-03-24 | Data enhancement method and system for improving crying detection model effect of baby |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113450776A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114333898A (en) * | 2021-12-10 | 2022-04-12 | 科大讯飞股份有限公司 | A sound event detection method, device, system and readable storage medium |
| CN114360523A (en) * | 2022-03-21 | 2022-04-15 | 深圳亿智时代科技有限公司 | Keyword dataset acquisition and model training methods, devices, equipment and medium |
| WO2025050263A1 (en) * | 2023-09-04 | 2025-03-13 | 南方科技大学 | Environment-adaptive baby cry detection method and system, storage medium and device |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| NL7103353A (en) * | 1970-03-13 | 1971-09-15 | ||
| JPH06124349A (en) * | 1992-10-12 | 1994-05-06 | Fujitsu Ltd | Pattern learning method and pattern learning device |
| KR20090119664A (en) * | 2008-05-16 | 2009-11-19 | 삼성전자주식회사 | System and method for object detection and classification with multiple threshold adaptive boosting |
| CN105139869A (en) * | 2015-07-27 | 2015-12-09 | 安徽清新互联信息科技有限公司 | Baby crying detection method based on interval difference features |
| KR20180120340A (en) * | 2017-04-27 | 2018-11-06 | 정재효 | System, method and program for analyzing the crying sound of baby |
| CN109902805A (en) * | 2019-02-22 | 2019-06-18 | 清华大学 | Deep Metric Learning and Apparatus for Adaptive Sample Synthesis |
| CN110085216A (en) * | 2018-01-23 | 2019-08-02 | 中国科学院声学研究所 | A kind of vagitus detection method and device |
| CN110363231A (en) * | 2019-06-27 | 2019-10-22 | 平安科技（深圳）有限公司 | Abnormality recognition method, device and storage medium based on semi-supervised deep learning |
-
2020
- 2020-03-24 CN CN202010215286.0A patent/CN113450776A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| NL7103353A (en) * | 1970-03-13 | 1971-09-15 | ||
| JPH06124349A (en) * | 1992-10-12 | 1994-05-06 | Fujitsu Ltd | Pattern learning method and pattern learning device |
| KR20090119664A (en) * | 2008-05-16 | 2009-11-19 | 삼성전자주식회사 | System and method for object detection and classification with multiple threshold adaptive boosting |
| CN105139869A (en) * | 2015-07-27 | 2015-12-09 | 安徽清新互联信息科技有限公司 | Baby crying detection method based on interval difference features |
| KR20180120340A (en) * | 2017-04-27 | 2018-11-06 | 정재효 | System, method and program for analyzing the crying sound of baby |
| CN110085216A (en) * | 2018-01-23 | 2019-08-02 | 中国科学院声学研究所 | A kind of vagitus detection method and device |
| CN109902805A (en) * | 2019-02-22 | 2019-06-18 | 清华大学 | Deep Metric Learning and Apparatus for Adaptive Sample Synthesis |
| CN110363231A (en) * | 2019-06-27 | 2019-10-22 | 平安科技（深圳）有限公司 | Abnormality recognition method, device and storage medium based on semi-supervised deep learning |
Non-Patent Citations (3)
| Title |
|---|
| NITESH V. CHAWLA等: "SMOTE:Synthetic Minority Over-sampling Technique", JOURAL OF ARTIFICIAL INTELLIGENCE RESEARCH, vol. 16, pages 321 - 357, XP055096179 * |
| 向鸿鑫等: "不平衡数据挖掘方法综述", 计算机工程与应用, vol. 55, no. 04 * |
| 赵宏旭;杨文帅;: "基于短时能量和梅尔倒谱系数的车型音频识别", 科学技术与工程, no. 18 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114333898A (en) * | 2021-12-10 | 2022-04-12 | 科大讯飞股份有限公司 | A sound event detection method, device, system and readable storage medium |
| CN114360523A (en) * | 2022-03-21 | 2022-04-15 | 深圳亿智时代科技有限公司 | Keyword dataset acquisition and model training methods, devices, equipment and medium |
| WO2025050263A1 (en) * | 2023-09-04 | 2025-03-13 | 南方科技大学 | Environment-adaptive baby cry detection method and system, storage medium and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113450776A (en) | Data enhancement method and system for improving crying detection model effect of baby | |
| Oster et al. | " Recognition" of Emotional Expression | |
| Leppänen et al. | Positive facial expressions are recognized faster than negative facial expressions, but why? | |
| Kuhl | Infants' Perception and Representation of Speech: Development of a New Theory. | |
| Lemaitre et al. | Vocal imitations and the identification of sound events | |
| Kätsyri | Those virtual people all look the same to me: Computer-rendered faces elicit a higher false alarm rate than real human faces in a recognition memory task | |
| Kohli et al. | The role of intelligent technologies in early detection of autism spectrum disorder (asd): A scoping review | |
| CN112906617A (en) | Driver abnormal behavior identification method and system based on hand detection | |
| Reggia et al. | What are the computational correlates of consciousness? | |
| CN113420745A (en) | Image-based target identification method, system, storage medium and terminal equipment | |
| CN114970701A (en) | Multi-mode fusion-based classroom interaction analysis method and system | |
| CN110472673B (en) | Parameter adjustment method, fundus image processing device, fundus image processing medium and fundus image processing apparatus | |
| CN119513569B (en) | Multi-granularity extraction and enhancement method, device and equipment for multi-source heterogeneous data | |
| Ahmed et al. | Visual deepfake detection: Review of techniques, tools, limitations, and future prospects | |
| CN115618298A (en) | Multi-modal model based 5G harmful message analysis method and device and storage medium | |
| CN114372906A (en) | Autism and other developmental disorder child teaching environment feedback obtaining method and device | |
| Huang et al. | Deep convolutional neural networks for the classification and detection of human vocal exclamations of panic in subway systems | |
| Gupta et al. | REDE-Detecting human emotions using CNN and RASA | |
| Yao et al. | A feature filter for EEG using Cycle-GAN structure | |
| CN116978125A (en) | Smoking detection method, device, electronic equipment and medium | |
| Kavitha et al. | Framework for Detecting Student Behaviour (Nail Biting, Sleep, and Yawn) Using Deep Learning Algorithm | |
| Chang et al. | Sensing with contexts: crying reason classification for infant care center with environmental fusion | |
| Sadasivam et al. | Multimodal Approach to Identify Attention Level of Students using Jetson Nano | |
| Mah et al. | Analysis of Virtual Reality Based on the Internet of Things on Human Psychology ‘Internet of Thoughts’(IoThs) for Rich Content Extraction Applied Natural Language Processing and Deep Learning | |
| CN112528790A (en) | Teaching management method and device based on behavior recognition and server |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210928 |
|
| RJ01 | Rejection of invention patent application after publication |