CN109559758A - A method of texture image is converted by haptic signal based on deep learning - Google Patents
A method of texture image is converted by haptic signal based on deep learning Download PDFInfo
- Publication number
- CN109559758A CN109559758A CN201811306545.XA CN201811306545A CN109559758A CN 109559758 A CN109559758 A CN 109559758A CN 201811306545 A CN201811306545 A CN 201811306545A CN 109559758 A CN109559758 A CN 109559758A
- Authority
- CN
- China
- Prior art keywords
- texture image
- tactile
- image
- spectrogram
- texture
- 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 27
- 238000013135 deep learning Methods 0.000 title claims abstract description 9
- 238000012549 training Methods 0.000 claims abstract description 22
- 238000001228 spectrum Methods 0.000 claims abstract description 18
- 230000001133 acceleration Effects 0.000 claims abstract description 9
- 238000010586 diagram Methods 0.000 claims description 13
- 230000005236 sound signal Effects 0.000 claims description 11
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 4
- 230000003321 amplification Effects 0.000 claims description 3
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 3
- 239000000463 material Substances 0.000 abstract description 11
- 230000003595 spectral effect Effects 0.000 abstract description 5
- 238000013473 artificial intelligence Methods 0.000 abstract description 3
- 238000012545 processing Methods 0.000 abstract description 3
- 230000009466 transformation Effects 0.000 abstract description 3
- 238000006243 chemical reaction Methods 0.000 abstract description 2
- 238000013256 Gubra-Amylin NASH model Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000004033 plastic Substances 0.000 description 1
- 229920003023 plastic Polymers 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 238000009877 rendering Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- -1 wooden Substances 0.000 description 1
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
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/06—Transformation of speech into a non-audible representation, e.g. speech visualisation or speech processing for tactile aids
- G10L21/16—Transforming into a non-visible representation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/40—Analysis of texture
- G06T7/41—Analysis of texture based on statistical description of texture
- G06T7/42—Analysis of texture based on statistical description of texture using transform domain methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Otolaryngology (AREA)
- Computational Linguistics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Signal Processing (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The method that texture image is converted into haptic signal based on deep learning that the present invention relates to a kind of, belongs to artificial intelligence, signal processing technology field.Learning training texture image data first obtains the characteristic information of image, to classify to all kinds of texture recognitions;The 3-axis acceleration signal of material surface friction vibration is converted into spectral image using Fourier algorithm in short-term, then training obtains spectral generator;Classification information is combined with spectral generator, automatically generates the frequency spectrum of texture image, frequency spectrum is converted into the haptic signal of different classes of image, realizes the conversion of different texture image to haptic signal.Result is passed into palm by being linked into the touch feedback device inside mouse, region locating for mouse pointer is tested texture area, so that sliding mouse can feed back the material properties for understanding testee in real time.The true sense of touch similarity of transformation result and image texture of the invention is higher, and application scenarios are abundant, has high practical value.
Description
Technical field
The method that texture image is converted into haptic signal based on deep learning that the present invention relates to a kind of, belongs to artificial intelligence
Energy, signal processing technology field.
Background technique
With the development of global industry and the emergence of artificial intelligence, object Material Identification is in e-commerce, mechanical system
It makes and is widely used with many industrial circles such as intelligent robot.Current Material Identification is typically based on the texture of body surface
Image identifies the affiliated material (such as wooden, glass, plastics, steel and fiber) of objects in images.But it is based on texture maps
The Material Identification of picture is easy to be influenced by shooting environmental, and apparent difference apparent between difference and small class is usual in big class
Will lead to textural characteristics can distinction weaken, robustness reduce.In addition to this, texture image can not accurate response and material phase
The thingness of pass.Such as, it is based on texture image, the textures and material object of identical texture cannot be distinguished.
For different texture images, the vibrational feedback signal under corresponding tool interaction can allow people preferably to sense it
Tactile characteristics.There are decades so far about touch feeling model building research, commonly entering in the modeling of tactile is tool
State (such as speed of tool) and grain surface state (such as attribute of texture), output be then vibrating tactile signal.
The limitation of complex mappings and experimental tool level between being output and input due to haptic signal, by the state and line of tool
The state for managing imaging surface is difficult as input simultaneously, can feed back its correspondence well there is no a kind of experimental model at present
Haptic signal.In addition to this, the quality for generating result for tactile is evaluated, and is remained in artificial subjective feeling at present and is differentiated
On, there is no the appraisement systems of a set of objective quantification, this allows for none good judgment criteria of result of study.
Summary of the invention
The method that texture image is converted into haptic signal based on deep learning that the purpose of the present invention is to propose to a kind of, with
The shortcoming for overcoming prior art, the haptic signal for being converted to texture image are more nearly true sense of touch.
Method that texture image is converted into haptic signal proposed by the present invention based on deep learning, including following step
It is rapid:
(1) Adam's optimization is carried out to the depth residual error network for texture image training, the depth optimized using the Adam
Residual error network is trained texture image, obtains the label information C with texture image feature;
(2) tactile frequency spectrum diagram generator is obtained using the 3-axis acceleration signal of texture image, process is as follows:
(2-1) carries out Short Time Fourier Transform to 3-axis acceleration signal relevant to texture image, obtains texture image
Initial haptic spectrogram, which is carried out taking logarithm and normalized, obtains the tactile frequency spectrum of texture image
Figure;
(2-2) obtains a tactile frequency spectrum with the tactile spectrogram of above-mentioned texture image, one confrontation learning network of training
Diagram generator G;
(3) to the label information C of step (2) tactile frequency spectrum diagram generator G input one random signal Z and step (1), touching
Feel that frequency spectrum diagram generator G exports tactile spectrogram corresponding with texture image;
(4) it by Griffin (Griffin-Lim) method, is generated and texture image phase by the tactile spectrogram of step (3)
The audio signal of generation is carried out power amplification by corresponding audio signal, and amplified audio signal is obtained through touch feedback device
The haptic signal of texture image.
Method that texture image is converted into haptic signal proposed by the present invention based on deep learning, its advantage is that:
The method of the present invention obtains the characteristic information of image using resnet model learning training texture image data, thus
Classify to all kinds of texture recognitions;Using Fourier algorithm in short-term, the 3-axis acceleration signal of each material surface friction vibration is turned
Change spectral image into;Followed by DC-GAN training, these data obtain spectral generator;It is generated using classification information and frequency spectrum
Device, realization automatically generate the frequency spectrum of different texture image, are then turned the spectrum results of generation using Griffin-Lim algorithm
It changes the haptic signal of different classes of image into, realizes the conversion of different texture image to haptic signal.Final result is by connecing
The touch feedback device entered to inside mouse passes to palm, and region locating for mouse pointer is tested texture area, so that
Sliding mouse can feed back the material properties for understanding testee in real time.The transformation result of the method for the present invention and image texture
True sense of touch similarity is higher, and application scenarios are abundant, has high practical value.
Detailed description of the invention
Fig. 1 is the flow diagram of the method for the present invention.
Fig. 2 is GAN model involved in the method for the present invention.
Specific embodiment
Method that texture image is converted into haptic signal proposed by the present invention based on deep learning, flow diagram is such as
Shown in Fig. 1, comprising the following steps:
(1) Adam's optimization is carried out to the depth residual error network for texture image training, the depth optimized using the Adam
Residual error network is trained texture image, obtains the label information C with texture image feature;
(2) tactile frequency spectrum diagram generator is obtained using the 3-axis acceleration signal of texture image, process is as follows:
(2-1) carries out Short Time Fourier Transform to 3-axis acceleration signal relevant to texture image, obtains texture image
Initial haptic spectrogram, which is carried out taking logarithm and normalized, obtains the tactile frequency spectrum of texture image
Figure;
The tactile spectrogram of (2-2) with above-mentioned texture image, one confrontation learning network of training, training process model such as figure
Shown in 2, a tactile frequency spectrum diagram generator G is obtained;
(3) to the label information C of step (2) tactile frequency spectrum diagram generator G input one random signal Z and step (1), touching
Feel that frequency spectrum diagram generator G exports tactile spectrogram corresponding with texture image;
(4) it by Griffin (Griffin-Lim) method, is generated and texture image phase by the tactile spectrogram of step (3)
The audio signal of generation is carried out power amplification by corresponding audio signal, and amplified audio signal is obtained through touch feedback device
The haptic signal of texture image.
Below in conjunction with attached drawing, the contents of the present invention are discussed in detail:
As shown in Figure 1, the present invention the following steps are included:
The extraction of characteristics of image is classified:
Pretreatment: due to the difference of all kinds of initial data itself, it is possible that the problems such as sample is less, distracter is more.
The operation for needing to carry out texture image data cleaning and data enhancing (mirror image, rotation, scaling, random cropping), then passes through
Screening, which is rejected, to be obtained being suitable for trained texture image data.
Resnet network is constructed using skip connection residual error module, is made it possible to training and is obtained deeper time
Network, to obtain preferable as a result, the present invention uses depth residual error network Resnet as training network, tool in this step
The model depth of body can adjust according to different results.Meanwhile the network optimized approach used in this step network is most
Adam Adam optimization.Faster, learning effect is more effective for Adam convergence speed of the algorithm, and can correct other optimisation techniques
The problem of, such as learning rate disappear, restrained slow or high variance parameter update cause loss function fluctuation it is larger
Problem.Some other specific parameter of prototype network can be adjusted according to different situations.After the completion of final mask training i.e.
The label information C of texture image can be obtained.
Spectrogram is converted by feature:
Corresponding tactile frequency is obtained by Short Time Fourier Transform (STFT) first with the acceleration signal of raw data base
Spectrogram, STFT (short-time Fourier transform, Short Time Fourier Transform) is relevant with Fourier transformation one
Kind mathematic(al) manipulation, to determine the frequency and phase of its regional area sine wave of time varying signal.Through STFT treated signal tool
There is the localization property of time domain and frequency domain, the consequential signal after Fourier transform in short-term has been carried out taking logarithm and returned in this step
One change processing, then generates to obtain desired spectrogram according to these information, parameter used in other Short Time Fourier Transforms
It can be adjusted according to different actual conditions.
Operation above only obtains training data, and the purpose of this step be autonomously generated with reference to training data it is desired
Spectrogram.Confrontation model, i.e. GAN model are inspired from two game sides in the zero-sum two-person game in game theory, GAN model point
It is not production model (generative model) and discriminative model (discriminative model).Generate model G
The distribution of sample data is captured, discrimination model D is two classifiers, estimates that a sample is (rather than raw from training data
At data) probability.It can make G by the game training of G and D, ultimately generate image information.
It is DC-GAN that the present invention, which generates network using confrontation, and DCGAN is relatively good improvement after GAN, greatly
Improve GAN training stability and generate outcome quality.For training data, all individually training is primary for each major class,
Such benefit is that the diversity that ensure that in class in turn ensures otherness between class.
DC-GAN model is as shown in Figure 2.DC-GAN network is mainly made of discriminator D and generator G two parts.Generator
Input be that C and Z are formed, C is the image vector that label information is had obtained in preceding step, and Z is random noise (according to difference
The case where dimensional parameter is set).Discriminator D first passes through the segments sonogram data collection that initial data obtains and is trained, and foundation is commented
Sentence index system, the spectrogram for then obtaining generator is as input, to complete to the differentiation for generating sonograph and return
Class.
Convert tactile:
It converts the C that texture image label information is had obtained in the extraction classifying step of characteristics of image to as feature
The input of the generation model for the DC-GAN that training finishes in spectrogram step, the two, which is combined, can be autonomously generated inhomogeneity
The corresponding spectrogram of other texture image.
The sonograph generated in previous step is generated into corresponding audio signal using Griffin (Griffin-Lim) algorithm,
Griffin-Lim algorithm can produce the audio signal of generation well according to frequency spectrum come reconstruction signal after power amplifier
Raw touch feedback, final result pass to palm, area locating for mouse pointer by being linked into the touch feedback device inside mouse
Domain is tested texture area, so that sliding mouse can feed back the material properties for understanding testee in real time.
It is the process evaluated the haptic signal generated using the method for the present invention below, calculating process is as follows:
The frequency of the original haptic signal by Texture image synthesis is sampled first, and draws waveform, extracts waveform
The coordinate of figure center line is (x1i, y1i);
Then waveform diagram is gone out by sample rendering to the haptic signal exported after generating model, extracts waveform diagram center
The coordinate of line is (x2i, y2i)。
Set similarity are as follows:
The value range of S is [0,1], and the tactile of the method for the present invention generation can be judged by the similarity of sampled signal
Signal quality, the more big then similarity of S value are higher.
Claims (1)
1. a kind of method that texture image is converted into haptic signal based on deep learning, it is characterised in that this method include with
Lower step:
(1) Adam's optimization is carried out to the depth residual error network for texture image training, the depth residual error optimized using the Adam
Network is trained texture image, obtains the label information C with texture image feature;
(2) tactile frequency spectrum diagram generator is obtained using the 3-axis acceleration signal of texture image, process is as follows:
(2-1) carries out Short Time Fourier Transform to 3-axis acceleration signal relevant to texture image, obtains the first of texture image
Beginning tactile spectrogram carries out taking logarithm and normalized to the tactile spectrogram, obtains the tactile spectrogram of texture image;
The tactile spectrogram of (2-2) with above-mentioned texture image, one confrontation learning network of training obtain a tactile spectrogram life
Grow up to be a useful person G;
(3) to the label information C of step (2) tactile frequency spectrum diagram generator G input one random signal Z and step (1), tactile frequency
Spectrogram generator G exports tactile spectrogram corresponding with texture image;
(4) by Griffin method, audio signal corresponding with texture image is generated by the tactile spectrogram of step (3), it will
The audio signal of generation carries out power amplification, and amplified audio signal obtains the tactile letter of texture image through touch feedback device
Number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811306545.XA CN109559758B (en) | 2018-11-05 | 2018-11-05 | Method for converting texture image into tactile signal based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811306545.XA CN109559758B (en) | 2018-11-05 | 2018-11-05 | Method for converting texture image into tactile signal based on deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109559758A true CN109559758A (en) | 2019-04-02 |
CN109559758B CN109559758B (en) | 2023-01-24 |
Family
ID=65865724
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811306545.XA Active CN109559758B (en) | 2018-11-05 | 2018-11-05 | Method for converting texture image into tactile signal based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109559758B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111796709A (en) * | 2020-06-02 | 2020-10-20 | 南京信息工程大学 | Method for reproducing image texture features on touch screen |
CN111796710A (en) * | 2020-06-02 | 2020-10-20 | 南京信息工程大学 | Method for reproducing image contour characteristics on touch screen |
CN112631434A (en) * | 2021-01-11 | 2021-04-09 | 福州大学 | Deep learning-based vibrotactile coding and decoding method |
CN113642604A (en) * | 2021-07-09 | 2021-11-12 | 南京邮电大学 | Audio and video auxiliary tactile signal reconstruction method based on cloud edge cooperation |
CN114265503A (en) * | 2021-12-22 | 2022-04-01 | 吉林大学 | Texture rendering method applied to pen type vibration tactile feedback device |
CN115758107A (en) * | 2022-10-28 | 2023-03-07 | 中国电信股份有限公司 | Haptic signal transmission method and apparatus, storage medium, and electronic device |
WO2024055416A1 (en) * | 2022-09-13 | 2024-03-21 | 瑞声开泰声学科技(上海)有限公司 | Haptic feedback signal generation method, electronic device, and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110244919A1 (en) * | 2010-03-19 | 2011-10-06 | Aller Joshua V | Methods and Systems for Determining Image Processing Operations Relevant to Particular Imagery |
CN104423592A (en) * | 2013-09-06 | 2015-03-18 | 意美森公司 | Systems and methods for generating haptic effects associated with an envelope in audio signals |
JP2018032236A (en) * | 2016-08-25 | 2018-03-01 | 株式会社ファセテラピー | Touch-feeling content generation device and touch-feeling content generation method |
GB201802645D0 (en) * | 2018-02-19 | 2018-04-04 | Valkyrie Industries Ltd | Haptic feedback for virtual reality |
CN108536780A (en) * | 2018-03-29 | 2018-09-14 | 清华大学 | A kind of cross-module state object material search method based on touch texture feature |
-
2018
- 2018-11-05 CN CN201811306545.XA patent/CN109559758B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110244919A1 (en) * | 2010-03-19 | 2011-10-06 | Aller Joshua V | Methods and Systems for Determining Image Processing Operations Relevant to Particular Imagery |
CN104423592A (en) * | 2013-09-06 | 2015-03-18 | 意美森公司 | Systems and methods for generating haptic effects associated with an envelope in audio signals |
JP2018032236A (en) * | 2016-08-25 | 2018-03-01 | 株式会社ファセテラピー | Touch-feeling content generation device and touch-feeling content generation method |
GB201802645D0 (en) * | 2018-02-19 | 2018-04-04 | Valkyrie Industries Ltd | Haptic feedback for virtual reality |
CN108536780A (en) * | 2018-03-29 | 2018-09-14 | 清华大学 | A kind of cross-module state object material search method based on touch texture feature |
Non-Patent Citations (3)
Title |
---|
HAITIAN ZHENG 等: "Deep Learning for Surface Material Classification Using Haptic and Visual Information", 《IEEE TRANSACTIONS ON MULTIMEDIA》 * |
张学军等: "基于图像处理的物体质量与软硬度的测量", 《广西大学学报(自然科学版)》 * |
李佳璐等: "彩色图像的纹理力/触觉渲染方法", 《计算机辅助设计与图形学学报》 * |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111796710B (en) * | 2020-06-02 | 2023-05-23 | 南京信息工程大学 | Method for reproducing image contour features on touch screen |
CN111796710A (en) * | 2020-06-02 | 2020-10-20 | 南京信息工程大学 | Method for reproducing image contour characteristics on touch screen |
CN111796709A (en) * | 2020-06-02 | 2020-10-20 | 南京信息工程大学 | Method for reproducing image texture features on touch screen |
CN111796709B (en) * | 2020-06-02 | 2023-05-26 | 南京信息工程大学 | Method for reproducing image texture features on touch screen |
CN112631434A (en) * | 2021-01-11 | 2021-04-09 | 福州大学 | Deep learning-based vibrotactile coding and decoding method |
CN112631434B (en) * | 2021-01-11 | 2022-04-12 | 福州大学 | Deep learning-based vibrotactile coding and decoding method |
CN113642604A (en) * | 2021-07-09 | 2021-11-12 | 南京邮电大学 | Audio and video auxiliary tactile signal reconstruction method based on cloud edge cooperation |
WO2023280064A1 (en) * | 2021-07-09 | 2023-01-12 | 南京邮电大学 | Audiovisual secondary haptic signal reconstruction method based on cloud-edge collaboration |
CN113642604B (en) * | 2021-07-09 | 2023-08-18 | 南京邮电大学 | Audio-video auxiliary touch signal reconstruction method based on cloud edge cooperation |
US11928957B2 (en) | 2021-07-09 | 2024-03-12 | Nanjing University Of Posts And Telecommunications | Audiovisual secondary haptic signal reconstruction method based on cloud-edge collaboration |
CN114265503A (en) * | 2021-12-22 | 2022-04-01 | 吉林大学 | Texture rendering method applied to pen type vibration tactile feedback device |
CN114265503B (en) * | 2021-12-22 | 2023-10-13 | 吉林大学 | Texture rendering method applied to pen-type vibration touch feedback device |
WO2024055416A1 (en) * | 2022-09-13 | 2024-03-21 | 瑞声开泰声学科技(上海)有限公司 | Haptic feedback signal generation method, electronic device, and storage medium |
CN115758107A (en) * | 2022-10-28 | 2023-03-07 | 中国电信股份有限公司 | Haptic signal transmission method and apparatus, storage medium, and electronic device |
CN115758107B (en) * | 2022-10-28 | 2023-11-14 | 中国电信股份有限公司 | Haptic signal transmission method and device, storage medium and electronic equipment |
Also Published As
Publication number | Publication date |
---|---|
CN109559758B (en) | 2023-01-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109559758A (en) | A method of texture image is converted by haptic signal based on deep learning | |
Demir et al. | A new deep CNN model for environmental sound classification | |
CN107463952B (en) | Object material classification method based on multi-mode fusion deep learning | |
Ullo et al. | Hybrid computerized method for environmental sound classification | |
CN101751692B (en) | Method for voice-driven lip animation | |
Wang et al. | Sound event recognition using auditory-receptive-field binary pattern and hierarchical-diving deep belief network | |
CN108681737B (en) | Method for extracting image features under complex illumination | |
Schramm et al. | Dynamic time warping for music conducting gestures evaluation | |
Miao et al. | Underwater acoustic signal classification based on sparse time–frequency representation and deep learning | |
Shinde et al. | Real time two way communication approach for hearing impaired and dumb person based on image processing | |
CN110490915A (en) | A kind of point cloud registration method being limited Boltzmann machine based on convolution | |
Zhao et al. | A survey on automatic emotion recognition using audio big data and deep learning architectures | |
Qais et al. | Deepfake audio detection with neural networks using audio features | |
CN116561533B (en) | Emotion evolution method and terminal for virtual avatar in educational element universe | |
CN109389101A (en) | A kind of SAR image target recognition method based on denoising autoencoder network | |
Wang et al. | Accelerometer-based gesture recognition using dynamic time warping and sparse representation | |
Abinaya | Acoustic based scene event identification using deep learning cnn | |
Godbole et al. | Indian language identification using deep learning | |
Meng et al. | Image-based handwriting analysis for disease diagnosis | |
CN114052675A (en) | Pulse anomaly distinguishing method and system based on fusion attention mechanism | |
Kumar et al. | Semantic segmentation of color images via feature extraction techniques | |
Khairina et al. | Preserving Cultural Heritage Through AI: Developing LeNet Architecture for Wayang Image Classification | |
Saranya et al. | AI based speech recognition of literacy to improve tribal English knowledge | |
Zeng et al. | Underwater image target detection with cascade classifier and image preprocessing method | |
Qian et al. | Target Classification in Unattended Ground Sensors With a Two-Stream Convolutional Network |
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 | ||
GR01 | Patent grant |