CN111860278A - Human behavior recognition algorithm based on deep learning - Google Patents
Human behavior recognition algorithm based on deep learning Download PDFInfo
- Publication number
- CN111860278A CN111860278A CN202010676134.0A CN202010676134A CN111860278A CN 111860278 A CN111860278 A CN 111860278A CN 202010676134 A CN202010676134 A CN 202010676134A CN 111860278 A CN111860278 A CN 111860278A
- Authority
- CN
- China
- Prior art keywords
- sample
- loss
- predicted
- data set
- behavior recognition
- 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
- 230000006399 behavior Effects 0.000 title claims abstract description 28
- 238000013135 deep learning Methods 0.000 title claims abstract description 16
- 230000006870 function Effects 0.000 claims abstract description 13
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000000034 method Methods 0.000 claims description 12
- 238000005070 sampling Methods 0.000 claims description 2
- 230000001419 dependent effect Effects 0.000 abstract 1
- 230000035945 sensitivity Effects 0.000 abstract 1
- 238000013527 convolutional neural network Methods 0.000 description 5
- 238000011176 pooling Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- QLMNCUHSDAGQGT-UHFFFAOYSA-N sintofen Chemical group N1=C(C(O)=O)C(=O)C=2C(OCCOC)=CC=CC=2N1C1=CC=C(Cl)C=C1 QLMNCUHSDAGQGT-UHFFFAOYSA-N 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- 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
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a human behavior recognition algorithm based on deep learning, which comprises the following steps of (1) preprocessing an input video segment; (2) constructing a network model RD 3D; (3) defining a loss function, an accuracy rate and an operation of an optimizer; (4) training the network model comprises the following substeps: (41) initializing parameters; (42) the learning rate is 0.0001, and the batch size is 16; (43) calculating loss by forward propagation values and real labels of the RD3D model, and updating weight parameters of the loss through backward propagation; (44) finishing training after 100 epochs are trained; (5) and (6) testing results. The invention pursues the accuracy of the recognition algorithm from the characteristic angle, overcomes the problem that the current algorithm is seriously dependent on the data set, reduces the sensitivity to the type of the data set, and can be applied to any behavior recognition data set.
Description
Technical Field
The invention relates to the technical field of computer vision, in particular to a human behavior recognition algorithm based on deep learning.
Background
In recent years, with the rise of related technologies such as deep learning, deep neural networks have made breakthrough progress in various fields such as computer vision. Deep learning is capable of learning its commonality features from training data due to its end-to-end training characteristics, and fitting a network suitable for the current task. Meanwhile, the mass data acquisition in the modern society becomes very easy, and convenience is provided for the application of deep learning to the fields of video understanding, identification and the like.
While in the traditional method, local features (such as HOG, HOF and MBH) are mainly extracted, and strong prior knowledge is required. Although the appearance and motion information are considered, the information is limited to a single frame, wherein the contextual appearance and motion information of the frame are ignored, which results in inaccurate human behavior recognition. How to design an algorithm for behavior recognition becomes particularly important.
Therefore, deep learning applications and human behavior recognition have been trending. The behavior recognition method based on deep learning mainly comprises the following steps: a dual-flow convolutional neural network, a 3D convolutional neural network, a convolutional neural network, and a cyclic neural combination. The invention improves the identification precision based on the 3D convolution network.
Patent CN 110163133 a "a human behavior recognition method based on depth residual error network" discloses a human behavior recognition method based on depth residual error network, which inputs human joint data and depth image data into ResNet for recognition at the same time, and although recognition accuracy is improved, human joint data and depth image are required as input, so end-to-end learning is impossible, and such data is lacking in daily life. Patent CN 107862275 a, "human behavior recognition model and its construction method and human behavior recognition method" discloses a method for recognizing human behavior by extracting human behavior feature vectors by using 3D convolutional neural network, inputting the extracted feature vectors into coulomb force field, and clustering by relatively moving all feature vectors under the action of the same class generating attraction and different classes generating repulsion. Inputting an RGB (red, green and blue) graph and an optical flow graph into a network for learning, the learning can not be carried out end to end, the whole network only has seven layers, only three layers carry out feature extraction, and although the calculation amount is small, the precision is low;
In the above, the recognition accuracy is improved from the perspective of conforming to a data set, and the accuracy of behavior recognition cannot be improved only by means of RGB images, and patent CN 109002808A, "a human behavior recognition method and system" discloses a human behavior recognition method and system, which train a 3D convolutional neural network by using a multi-task deep learning method, and complete a recognition task after training by using continuous video frames of various human behavior attributes and background videos as inputs. More, how to make a data set in the multitask learning is taught, so that behavior videos and background videos are distinguished, feature extraction is completed only by means of seven layers of common 3D convolutional networks, and classification is achieved. The identification of human behavior is still done from a dataset perspective.
Disclosure of Invention
Aiming at the technical problem, the invention provides a human behavior recognition algorithm based on deep learning, which comprises the following steps:
(1) preprocessing an input video segment;
(2) constructing a network model RD 3D;
(3) defining a loss function and an optimizer operation;
(4) training the network model comprises the following substeps:
(41) initializing parameters;
(42) the learning rate is 0.0001, and the batch size is 16;
(43) Calculating loss according to a loss function by the forward propagation value and the real label of the RD3D model, and updating a weight parameter of the loss through backward propagation;
(44) finishing training after 100 epochs are trained;
(5) and (6) testing results.
Further, in the preprocessing stage in the step (1), in order to comprehensively consider the global motion information of the video, a subsampling algorithm is provided and adopted to collect n frames of key video frames, so that the recognition accuracy is improved, and the specific content is as follows:
a: acquiring an image frame (alpha is 3) of each video clip according to an acquisition rate alpha to obtain an image data set A corresponding to each video;
b: uniformly collecting n frames (n is 16) from the image data set A by adopting a subsampling algorithm to serve as key frames of the video clip, and scaling the key frames to k is k (k is 224) to form a data set B;
d: the acquired data set B was assigned a 7: and 3, dividing the ratio into a training set and a testing set for training and testing, wherein each sample in the training set is an achor, positive, negative, label, and is respectively a sample to be predicted, other samples in the same class as the sample to be predicted, other samples in different classes as the sample to be predicted, and class labels of the samples to be predicted.
Further, in the step (2), in order to improve the identification accuracy, a novel network model RD3D (Residual Dense 3D) is proposed and designed by combining a feature multiplexing idea and a shortcut idea. The model RD3D designs 134 layers, namely 1+4 × 4+6 × 3+2 × 4+1, 6 stages.
Further, step (3) proposes and designs a novel loss function:
F=H(P,Q)+Lre+Ltr
wherein:
the cross entropy H (P, Q) ═ P (x) log (Q (x)), measures the similarity of the predicted distribution and the true distribution, and the smaller the loss, the more accurate the classification. Wherein P is the true sample distribution and Q is the predicted sample distribution;
l2 regularization lossTo prevent overfitting, λ is a penalty factor (λ ═ 0.009), and n is the number of weights W;
ternary lossWhereinIs xiAnd xp iThe Euclidean distance of (a) is,is xiAnd xn iF (x) is the feature of sample x extracted by RD3D, bs is batch size, xiFor the currently predicted sample, xp iCalculating a sample x for the sum currentiSamples of the same class, xn iCalculating a sample x for the sum currentiNot of the same class, beta is xiAnd xp i、xiAnd xn iIs equal to 0.2.
The invention overcomes the problem that the current algorithm depends on the data set seriously while pursuing the accuracy of the recognition algorithm, realizes the network structure design from the characteristic angle of human behavior extraction, is insensitive to the type of the data set, and can be applied to any data set.
Drawings
FIG. 1 is a RD3D model of the present invention;
FIG. 2 is a Conv Block structure according to the present invention;
FIG. 3 is an ID Block structure of the present invention;
FIG. 4 is a flow chart of the present invention.
Detailed Description
The specific technical scheme of the invention is described by combining the embodiment.
As shown in fig. 4, a human behavior recognition algorithm based on deep learning includes the following steps:
(1) preprocessing an input video segment (the embodiment takes a UCF101 data set as an example);
(2) constructing a network model RD 3D;
(3) defining a loss function, an accuracy rate and an operation of an optimizer;
(4) training the network model comprises the following substeps:
(41) initializing parameters;
(42) the learning rate is 0.0001, and the batch size is 16;
(43) calculating loss according to a loss function by the forward propagation value and the real label of the RD3D model, and updating a weight parameter of the loss through backward propagation;
(44) finishing training after 100 epochs are trained;
(5) and (6) testing results.
Specifically, the method comprises the following steps:
(1) in the preprocessing stage, in order to comprehensively consider the global motion information of the video, a secondary sampling algorithm is provided and adopted to collect n frames of key frames, so that the identification accuracy is improved, and the specific content is as follows:
a: acquiring an image frame (alpha is 3) of each video clip according to an acquisition rate alpha to obtain an image data set A corresponding to each video;
b: uniformly collecting n frames (n is 16) from the image data set A by adopting a subsampling algorithm to serve as key frames of the video clip, and scaling the key frames to k is k (k is 224) to form a data set B;
d: the acquired data set B was assigned a 7: and 3, dividing the ratio into a training set and a testing set for training and testing, wherein each sample in the training set is an achor, positive, negative, label, and is respectively a sample to be predicted, other samples in the same class as the sample to be predicted, other samples in different classes as the sample to be predicted, and class labels of the samples to be predicted.
(2) In order to improve the identification accuracy, a novel network model RD3D (Residual Dense3D) is proposed and designed by combining a feature multiplexing idea and a shortcut idea, the structure of the model is shown in FIG. 1, 127 layers are designed for the RD3D model, and the contents are as follows:
a: stage1 consists of Conv3d, BN, Relu, MaxPool, where Conv3d has 64 filters, convolution kernel 3 × 3, stride [1,2,2], padding is SAME; the pooling window in MaxPool is 1 × 3, stride ═ 1,2, 2. Stage1 has an input dimension of [16,16, 224, 224, 3], an output dimension of [16,16,56,56,64 ];
b: stage2 is composed of Conv Block4, three ID Block4, Maxpool, wherein Conv Block4 is composed of 4 layers of 3D convolution groups and shortcuts with two layers of convolution, which are connected together by channel addition, as shown in FIG. 2, wherein the filter numbers of the 4 layers of 3D convolution groups are 64, 64, 128, 128, respectively, and the convolution kernels are all 3 × 3; in a convolution group, the input of the next layer is the output of all previous layers in the block. The number of filters in shortcut is 128 and the convolution kernels are 1 x 1, 3 x 3, respectively. ID Block4 is formed by adding 4 layers of 3D convolution groups and the input of the Block, as shown in fig. 3, wherein the filter numbers of the 4 layers of 3D convolution groups are 64, 64, 128, 128, respectively, and the convolution kernels are all 3 × 3; in a convolution group, the input of the next layer is the output of all previous layers in the block. Pooling window 2 x 2, stride ═ 2,2,2 in Maxpool. Stage2 has an input dimension of [16,16,56,56,64], an output dimension of [16,8,28, 128 ];
C: stage3, stage4 are identical in composition with stage2, the only difference is that the number of layers and the number of filters in each block are different, in stage3, the number of layers of ConvBlock and IDBlock are all 6, the number of filters in each layer is 128, 128, 256, 256, 512, 512, 512, the number of filters of shortcut in ConvBlock is 512, the input dimension of stage3 is [16,8,28, 128], the output dimension is [16,4,14, 512 ]; in stage4, the number of layers of ConvBlock and IDBlock is 6, the number of filters in each layer is 256, 256, 512, 512, 1024, 1024, and the number of filters in shortcut in ConvBlock is 1024. Stage4 has an input dimension of [16,4,14, 512], an output dimension of [16,2,7, 1024 ];
d: stage5 and stage2 differ in composition in that stage5 has no MaxPool. In ConvBlock and IDBlock, the number of layers is 6, the number of filters in each layer is 512, 512, 1024, 1024, 2048, 2048, respectively, the number of filters for shortcut in ConvBlock is 2048, the input dimension of stage5 is [16,2,7, 1024], and the output dimension is [16,2,7, 2048 ];
e: stage6 is composed of AvgPool, platten, FC, Softmax, as shown in fig. 1, where AvgPool is global mean pooling, the pooling window is 2 × 7, platten is to set the output reshape of the upper layer to [16,2048], FC is a fully-connected layer, the output dimension is class number 101 of ucf101, and Softmax is a classified layer. Stage6 has an input dimension of [16,2,7, 2048] and an output dimension of [16,101]
(3) A loss function is designed. Conventional loss function F ═ H (P, Q) + LreIn order to enlarge the discrimination of different types of samples and improve the identification precision, the invention adds ternary loss on the basis of the traditional loss function to obtain a new loss function:
F=H(P,Q)+Lre+Ltrwherein the content of the first and second substances,
a: the cross entropy H (P, Q) ═ P (x) log (Q (x)), measures the similarity of the predicted distribution and the true distribution, and the smaller the loss, the more accurate the classification. Wherein P is the true sample distribution and Q is the predicted sample distribution;
b: l2 regularization lossTo prevent overfitting, λ is a penalty factor (λ ═ 0.009), and n is the number of weights W;
c: ternary loss:
whereinIs xiAnd xp iThe Euclidean distance of (a) is,is xiAnd xn iF (x) is the feature of sample x extracted by RD3D, bs is batch size, xiFor the currently predicted sample, xp iCalculating a sample x for the sum currentiSamples of the same class, xn iCalculating a sample x for the sum currentiNot of the same class, beta is xiAnd xp i、xiAnd xn iIs equal to 0.2.
Claims (4)
1. A human behavior recognition algorithm based on deep learning is characterized by comprising the following steps:
(1) preprocessing an input video segment;
(2) constructing a network model RD 3D;
(3) a loss function and the operation of the optimizer are defined.
(4) Training a network model; the method comprises the following substeps:
(41) initializing parameters;
(42) the learning rate is 0.0001, and the batch size is 16;
(43) calculating loss according to a loss function by the forward propagation value and the real label of the RD3D model, and updating a weight parameter of the loss through backward propagation;
(44) finishing training after 100 epochs are trained;
(5) and (6) testing results.
2. The deep learning-based human behavior recognition algorithm according to claim 1, wherein in the preprocessing stage of step (1), a subsampling algorithm is proposed and adopted to collect n frames of key frames, and the method specifically comprises the following steps:
a: acquiring image frames of each video clip according to an acquisition rate alpha to obtain an image data set A corresponding to each video;
b: uniformly collecting n frames from the image data set A by adopting a secondary sampling algorithm to serve as key frames of the video clips, and scaling the key frames to k x k to form a data set B;
d: the acquired data set B was assigned a 7: and 3, dividing the sample into a training set and a testing set in proportion for training and testing, wherein each sample in the training set is a quadruple and is respectively a sample to be predicted, other samples in the same class as the sample to be predicted, other samples in different classes as the sample to be predicted, and class labels of the sample to be predicted.
3. The deep learning-based human behavior recognition algorithm of claim 1, wherein the RD3D model of step (2) designs 134 layers, i.e. 1+ 4+6 + 3+2 + 4+1, 6 stages.
4. The deep learning-based human behavior recognition algorithm according to claim 1, wherein the step (3) designs a loss function:
F=H(P,Q)+Lre+Ltr
wherein, the cross entropy H (P, Q) ═ P (x) log (Q (x)) measures the similarity of the predicted distribution and the real distribution, and the smaller the loss is, the more accurate the classification is; wherein P is the true sample distribution and Q is the predicted sample distribution;
l2 regularization lossIn order to prevent overfitting, lambda is a penalty factor, and n is the number of weights W;
ternary loss:
wherein | | | f (x)i)-f(xp i)||2 2Is xiAnd xp iEuclidean distance of, | f (x)i)-f(xn i)||2 2Is xiAnd xn iF (x) is the feature of sample x extracted by RD3D, bs is batch size, xiFor the currently predicted sample, xp iCalculating a sample x for the sum currentiSamples of the same class, xn iCalculating a sample x for the sum currentiNot of the same class, beta is xiAnd xp i、xiAnd xn iIs measured.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010676134.0A CN111860278B (en) | 2020-07-14 | 2020-07-14 | Human behavior recognition algorithm based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010676134.0A CN111860278B (en) | 2020-07-14 | 2020-07-14 | Human behavior recognition algorithm based on deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111860278A true CN111860278A (en) | 2020-10-30 |
CN111860278B CN111860278B (en) | 2024-05-14 |
Family
ID=72984833
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010676134.0A Active CN111860278B (en) | 2020-07-14 | 2020-07-14 | Human behavior recognition algorithm based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111860278B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112861752A (en) * | 2021-02-23 | 2021-05-28 | 东北农业大学 | Crop disease identification method and system based on DCGAN and RDN |
CN113361417A (en) * | 2021-06-09 | 2021-09-07 | 陕西理工大学 | Human behavior identification method based on variable time sequence |
CN114897146A (en) * | 2022-05-18 | 2022-08-12 | 北京百度网讯科技有限公司 | Model generation method and device and electronic equipment |
Citations (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017133009A1 (en) * | 2016-02-04 | 2017-08-10 | 广州新节奏智能科技有限公司 | Method for positioning human joint using depth image of convolutional neural network |
WO2017203262A2 (en) * | 2016-05-25 | 2017-11-30 | Metail Limited | Method and system for predicting garment attributes using deep learning |
CN108009525A (en) * | 2017-12-25 | 2018-05-08 | 北京航空航天大学 | A kind of specific objective recognition methods over the ground of the unmanned plane based on convolutional neural networks |
CN108830185A (en) * | 2018-05-28 | 2018-11-16 | 四川瞳知科技有限公司 | Activity recognition and localization method based on multitask combination learning |
CN109993076A (en) * | 2019-03-18 | 2019-07-09 | 华南理工大学 | A kind of white mouse behavior classification method based on deep learning |
WO2019169942A1 (en) * | 2018-03-09 | 2019-09-12 | 华南理工大学 | Anti-angle and occlusion interference fast face recognition method |
CN110348381A (en) * | 2019-07-11 | 2019-10-18 | 电子科技大学 | A kind of video behavior recognition methods based on deep learning |
WO2019232894A1 (en) * | 2018-06-05 | 2019-12-12 | 中国石油大学(华东) | Complex scene-based human body key point detection system and method |
CN110598598A (en) * | 2019-08-30 | 2019-12-20 | 西安理工大学 | Double-current convolution neural network human behavior identification method based on finite sample set |
CN110619352A (en) * | 2019-08-22 | 2019-12-27 | 杭州电子科技大学 | Typical infrared target classification method based on deep convolutional neural network |
CN110826462A (en) * | 2019-10-31 | 2020-02-21 | 上海海事大学 | Human body behavior identification method of non-local double-current convolutional neural network model |
CN110956085A (en) * | 2019-10-22 | 2020-04-03 | 中山大学 | Human behavior recognition method based on deep learning |
WO2020073951A1 (en) * | 2018-10-10 | 2020-04-16 | 腾讯科技(深圳)有限公司 | Method and apparatus for training image recognition model, network device, and storage medium |
CN111027487A (en) * | 2019-12-11 | 2020-04-17 | 山东大学 | Behavior recognition system, method, medium, and apparatus based on multi-convolution kernel residual network |
-
2020
- 2020-07-14 CN CN202010676134.0A patent/CN111860278B/en active Active
Patent Citations (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017133009A1 (en) * | 2016-02-04 | 2017-08-10 | 广州新节奏智能科技有限公司 | Method for positioning human joint using depth image of convolutional neural network |
WO2017203262A2 (en) * | 2016-05-25 | 2017-11-30 | Metail Limited | Method and system for predicting garment attributes using deep learning |
CN108009525A (en) * | 2017-12-25 | 2018-05-08 | 北京航空航天大学 | A kind of specific objective recognition methods over the ground of the unmanned plane based on convolutional neural networks |
WO2019169942A1 (en) * | 2018-03-09 | 2019-09-12 | 华南理工大学 | Anti-angle and occlusion interference fast face recognition method |
CN108830185A (en) * | 2018-05-28 | 2018-11-16 | 四川瞳知科技有限公司 | Activity recognition and localization method based on multitask combination learning |
WO2019232894A1 (en) * | 2018-06-05 | 2019-12-12 | 中国石油大学(华东) | Complex scene-based human body key point detection system and method |
WO2020073951A1 (en) * | 2018-10-10 | 2020-04-16 | 腾讯科技(深圳)有限公司 | Method and apparatus for training image recognition model, network device, and storage medium |
CN109993076A (en) * | 2019-03-18 | 2019-07-09 | 华南理工大学 | A kind of white mouse behavior classification method based on deep learning |
CN110348381A (en) * | 2019-07-11 | 2019-10-18 | 电子科技大学 | A kind of video behavior recognition methods based on deep learning |
CN110619352A (en) * | 2019-08-22 | 2019-12-27 | 杭州电子科技大学 | Typical infrared target classification method based on deep convolutional neural network |
CN110598598A (en) * | 2019-08-30 | 2019-12-20 | 西安理工大学 | Double-current convolution neural network human behavior identification method based on finite sample set |
CN110956085A (en) * | 2019-10-22 | 2020-04-03 | 中山大学 | Human behavior recognition method based on deep learning |
CN110826462A (en) * | 2019-10-31 | 2020-02-21 | 上海海事大学 | Human body behavior identification method of non-local double-current convolutional neural network model |
CN111027487A (en) * | 2019-12-11 | 2020-04-17 | 山东大学 | Behavior recognition system, method, medium, and apparatus based on multi-convolution kernel residual network |
Non-Patent Citations (3)
Title |
---|
张延安;王宏玉;徐方;: "基于深度卷积神经网络与中心损失的人脸识别", 科学技术与工程, no. 35, 18 December 2017 (2017-12-18), pages 97 - 102 * |
解怀奇;乐红兵;: "基于通道注意力机制的视频人体行为识别", 电子技术与软件工程, no. 04, 15 February 2020 (2020-02-15), pages 146 - 148 * |
赵新秋;杨冬冬;贺海龙;段思雨;: "基于深度学习的人体行为识别研究", 高技术通讯, no. 05, 15 May 2020 (2020-05-15), pages 41 - 49 * |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112861752A (en) * | 2021-02-23 | 2021-05-28 | 东北农业大学 | Crop disease identification method and system based on DCGAN and RDN |
CN112861752B (en) * | 2021-02-23 | 2022-06-14 | 东北农业大学 | DCGAN and RDN-based crop disease identification method and system |
CN113361417A (en) * | 2021-06-09 | 2021-09-07 | 陕西理工大学 | Human behavior identification method based on variable time sequence |
CN113361417B (en) * | 2021-06-09 | 2023-10-31 | 陕西理工大学 | Human behavior recognition method based on variable time sequence |
CN114897146A (en) * | 2022-05-18 | 2022-08-12 | 北京百度网讯科技有限公司 | Model generation method and device and electronic equipment |
CN114897146B (en) * | 2022-05-18 | 2023-11-03 | 北京百度网讯科技有限公司 | Model generation method and device and electronic equipment |
Also Published As
Publication number | Publication date |
---|---|
CN111860278B (en) | 2024-05-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109840471B (en) | Feasible road segmentation method based on improved Unet network model | |
CN111738301B (en) | Long-tail distribution image data identification method based on double-channel learning | |
CN111860278B (en) | Human behavior recognition algorithm based on deep learning | |
CN112308158A (en) | Multi-source field self-adaptive model and method based on partial feature alignment | |
CN111144448A (en) | Video barrage emotion analysis method based on multi-scale attention convolutional coding network | |
CN109255284B (en) | Motion trajectory-based behavior identification method of 3D convolutional neural network | |
CN102385592B (en) | Image concept detection method and device | |
CN113221641A (en) | Video pedestrian re-identification method based on generation of confrontation network and attention mechanism | |
CN111652273B (en) | Deep learning-based RGB-D image classification method | |
CN110751027B (en) | Pedestrian re-identification method based on deep multi-instance learning | |
CN111461129B (en) | Context prior-based scene segmentation method and system | |
CN114067118B (en) | Processing method of aerial photogrammetry data | |
CN110728694A (en) | Long-term visual target tracking method based on continuous learning | |
CN112329536A (en) | Single-sample face recognition method based on alternative pair anti-migration learning | |
CN114038037A (en) | Expression label correction and identification method based on separable residual attention network | |
CN102184384A (en) | Face identification method based on multiscale local phase quantization characteristics | |
CN115170868A (en) | Clustering-based small sample image classification two-stage meta-learning method | |
CN108537109A (en) | Monocular camera sign Language Recognition Method based on OpenPose | |
CN113032613B (en) | Three-dimensional model retrieval method based on interactive attention convolution neural network | |
CN110889335A (en) | Human skeleton double-person interaction behavior recognition method based on multi-channel space-time fusion network | |
CN113989556A (en) | Small sample medical image classification method and system | |
CN109919057B (en) | Multi-mode fusion gesture recognition method based on efficient convolutional neural network | |
CN106250818B (en) | A kind of total order keeps the face age estimation method of projection | |
CN114898464B (en) | Lightweight accurate finger language intelligent algorithm identification method based on machine vision | |
CN115797827A (en) | ViT human body behavior identification method based on double-current network architecture |
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 |