CN112163487A - Gait identity recognition method based on improved space-time gait energy diagram - Google Patents
Gait identity recognition method based on improved space-time gait energy diagram Download PDFInfo
- Publication number
- CN112163487A CN112163487A CN202010994932.8A CN202010994932A CN112163487A CN 112163487 A CN112163487 A CN 112163487A CN 202010994932 A CN202010994932 A CN 202010994932A CN 112163487 A CN112163487 A CN 112163487A
- Authority
- CN
- China
- Prior art keywords
- gait
- time
- improved space
- recognition
- energy diagram
- 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
- 230000005021 gait Effects 0.000 title claims abstract description 199
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000010586 diagram Methods 0.000 title claims abstract description 29
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 21
- 230000008859 change Effects 0.000 claims abstract description 6
- 238000012549 training Methods 0.000 claims description 24
- 210000000544 articulatio talocruralis Anatomy 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 12
- 210000004394 hip joint Anatomy 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 7
- 210000001503 joint Anatomy 0.000 claims description 6
- 239000003086 colorant Substances 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 230000001351 cycling effect Effects 0.000 claims description 3
- 238000001514 detection method Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000004791 biological behavior Effects 0.000 abstract description 4
- 230000007547 defect Effects 0.000 abstract description 4
- 230000008569 process Effects 0.000 description 9
- 230000000694 effects Effects 0.000 description 4
- 238000005457 optimization Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000002194 synthesizing effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 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
- G06V40/23—Recognition of whole body movements, e.g. for sport training
- G06V40/25—Recognition of walking or running movements, e.g. gait 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
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Molecular Biology (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Psychiatry (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a gait identity recognition method based on an improved space-time gait energy diagram, which is characterized in that the improved space-time gait energy diagram is combined with a convolutional neural network GoogLeNet to establish a gait recognition model, the improved space-time gait energy diagram is taken as input to extract and classify gait recognition characteristics, the ruleless change and swing amplitude characteristics of the lower half body joints of a human body are extracted, time sequence information is mapped onto the lower half body joints by utilizing a pseudo-color coding technology, and the internal dynamic information of a gait template is reserved to the maximum extent, so that a richer and effective gait template is obtained, and the improved space-time gait energy diagram is generated. The defects of long gait recognition time based on a model, complex modeling and low cross-view angle accuracy based on image gait recognition are overcome, the modern biological behavior characteristic recognition technology is improved, the gait identity recognition time is reduced while the remote identity recognition safety is ensured, and a method guarantee is provided for the effective real-time identity recognition technology.
Description
Technical Field
The invention relates to the technical field of gait recognition, in particular to a gait identity recognition method based on an improved space-time gait energy diagram.
Background
Gait refers to the posture of people when walking, and is a biological behavior characteristic capable of being remotely sensed. Gait recognition is a technology for carrying out identity recognition according to gait differences of different individuals, and the method has the main advantages of realizing remote and hidden identity recognition and being widely applied to the field of intelligent video monitoring. At present, the existing gait recognition methods are divided into model-based and non-model-based methods. A model-based gait recognition method (GEI) includes modeling a static structure of a human body. The model-based method directly uses the static and motion parameters of the human body as gait characteristics, and can process occlusion and noise to a certain extent. However, the model-based method firstly needs to accurately position the position of the human body joint point, is easily affected by image quality, and has long and complicated modeling process, so that it is generally difficult to extract the essential model of gait movement. The gait recognition method based on the non-model mainly carries out feature analysis on the outline, the shape original image and the like of human gait so as to carry out identity recognition, and the gait recognition method based on the non-model is superior to the method based on the model in computational complexity because of no complex modeling process, but has obvious advantages in solving the challenges of gait recognition such as shielding, cross-view angle and the like.
Aiming at the problems of low accuracy rate under the cross-view angle based on non-model (image) gait recognition and low accuracy rate under the single-view angle based on model gait recognition, the invention provides a gait template based on an improved space-time gait energy map by combining a gait recognition method based on a model and a non-model. And gait recognition research is carried out by combining a deep learning method so as to improve the accuracy of cross-visual angle and cross-state gait recognition.
Disclosure of Invention
In order to solve some or some technical problems in the prior art, the invention provides a gait identification method based on an improved space-time gait energy diagram, which solves the defects of long gait identification time, complicated modeling and low cross-view angle accuracy based on model gait identification, improves the modern biological behavior characteristic identification technology, ensures the safety of remote identity identification, reduces the time required by gait identity identification, and provides method guarantee for the effective real-time identity identification technology.
In order to solve the above-mentioned existing technical problem, the invention adopts the following scheme:
a gait identity recognition method based on an improved space-time gait energy map is characterized in that the improved space-time gait energy map and a convolutional neural network GoogLeNet are combined to establish a gait recognition model, the improved space-time gait energy map is used as input to extract and classify gait recognition features, and the improved space-time gait energy map generation steps are as follows:
(1) preprocessing all gait contour maps in a gait database by image extraction, center standardization and the like;
(2) determining coordinates of a hip joint and two ankle joints of a given gait contour diagram by using an existing related algorithm, obtaining joint point positioning information, and extracting regular changes between joints of the lower half of a human body;
(3) connecting the distance between the hip joint and the left and right ankle joints to obtain the characteristics of regular change and swing amplitude between the joint points of the lower half body;
(4) combining a pseudo color coding technology with a stride, mapping time sequence information to the lower body joint points, and describing the time sequence information of a gait outline graph through pseudo colors;
(5) and (3) cycling all gait contour maps in the database through gait cycle detection, and averagely superposing the gait contour maps in the cycle after the gait contour maps are processed in the steps (1), (2) and (3) to obtain a new gait template containing time information and joint dynamic information.
Furthermore, the processed gait data set is divided into a plurality of sub-test sets and sub-training sets, a GoogleNe deep convolutional neural network is combined, the new gait templates divided into the sub-training sets are used as input to carry out gait recognition feature extraction training, and meanwhile, the new gait templates of the sub-test sets are used for verifying the effectiveness of the improved space-time gait energy diagram provided by the invention.
Further, the training set of the sub-training set is set as a gait sequence of 0-180 degrees in a normal state.
Further, the gait sequence angles set by the sub-training set do not include 0 °, 54 °, 90 ° and 126 °.
Further, the sub-training set is set to be in a gait sequence of 36 degrees to 144 degrees in a normal state.
Compared with the prior art, the invention has the beneficial effects that:
the invention extracts the regular change and swing characteristic between the joints of the lower half of the human body, utilizes the pseudo-color coding technology to map the time sequence information onto the joints of the lower half of the human body, retains the dynamic information in the gait template to the utmost extent, thereby obtaining a richer and more effective gait template, and generates a gait pattern (ICGI) based on the improved space-time gait energy, the method combines the convolutional neural network GoogleLeNet to establish a gait recognition model, compared with the gait recognition prediction model based on the traditional gait energy pattern and the space-time gait energy pattern, the provided method for recognizing the gait identity based on the improved space-time gait energy pattern has higher gait recognition accuracy, the invention solves the defects of long time and complicated modeling based on the model gait recognition, low cross-view angle accuracy based on the image gait recognition, improves the modern biological behavior characteristic recognition technology, ensures the remote identity recognition safety, the time required by gait identification is reduced, and a method guarantee is provided for an effective real-time identification technology.
Drawings
Fig. 1 is a schematic diagram of the generation of spatiotemporal gait energy diagrams.
Fig. 2 is a gait recognition model based on google lenet.
Detailed Description
The present invention will be further described with reference to the accompanying drawings and the detailed description, and it should be noted that any combination of the embodiments or technical features described below can be used to form a new embodiment without conflict.
As shown in fig. 1 and 2, a gait identification method based on an improved space-time gait energy diagram, which combines the improved space-time gait energy diagram with a convolutional neural network google lenet to establish a gait identification model, and uses the improved space-time gait energy diagram as input to extract and classify gait identification features, wherein the improved space-time gait energy diagram generation steps are as follows:
(1) preprocessing all gait contour maps in a gait database by image extraction, center standardization and the like;
(2) determining coordinates of a hip joint and two ankle joints of a given gait contour diagram by using an existing related algorithm, obtaining joint point positioning information, and extracting regular changes between joints of the lower half of a human body;
(3) connecting the distance between the hip joint and the left and right ankle joints to obtain the characteristics of regular change and swing amplitude between the joint points of the lower half body;
(4) combining a pseudo color coding technology with a stride, mapping time sequence information to the lower body joint points, and describing the time sequence information of a gait outline graph through pseudo colors;
(5) and (3) cycling all gait contour maps in the database through gait cycle detection, and averagely superposing the gait contour maps in the cycle after the gait contour maps are processed in the steps (1), (2) and (3) to obtain a new gait template containing time information and joint dynamic information.
The method is further improved in that the processed gait data set is divided into a plurality of sub-test sets and sub-training sets, a GoogleNe deep convolutional neural network is combined, the new gait templates divided into the sub-training sets are used as input to conduct gait recognition feature extraction training, and meanwhile the new gait templates of the sub-test sets are used for verifying the effectiveness of the improved space-time gait energy diagram.
Determining the coordinates of a hip joint of the lower half of a human body and a left ankle joint and a right ankle joint of the human body according to an existing related algorithm, adding color lines to the distance between joint points by using a pseudo color coding technology, extracting the regular change and swing characteristics between joints of the lower half of the human body, mapping time sequence information onto the distance between the joint points by using the pseudo color coding technology, reserving dynamic information in a Gait template to the maximum extent, synthesizing a single-frame Gait contour map processed in a period into an Image according to weighted average, thereby obtaining a richer and effective Gait template, generating a Gait recognition model based on an Improved space-time Gait energy map ICGI (Improved chord-Gait Image), and establishing the Gait recognition model by combining a convolutional neural network GoogleNet to extract and classify the Gait recognition characteristics by taking the Improved space-time Gait energy map as input. Compared with a gait recognition prediction model based on a traditional gait energy map and a space-time gait energy map, the gait identity recognition method based on the improved space-time gait energy map verifies the recognition effect by using the existing database, and the result shows that the gait identity recognition method based on the improved space-time gait energy map greatly improves the recognition effect compared with the traditional machine learning gait recognition method and has higher gait recognition accuracy.
The ICGI generation schematic diagram of the improved space-time gait energy diagram comprises the steps of firstly extracting a used gait data set, standardizing the center, and positioning the coordinates of the hip joint of the lower half of the human body and the coordinates of the left ankle joint and the right ankle joint of the human body by combining human engineering and related algorithms. By utilizing a pseudo-color coding technology and lower body joint point information, a single frame gait contour graph processed in a period is synthesized into an image, namely a gait space-time energy graph according to weighted average. Determining the coordinates of the hip joint, the left ankle joint and the right ankle joint of the lower half of the human body according to the existing related algorithm, adding the distance between joint points into a color line by using a pseudo color coding technology, synthesizing a single-frame Gait contour map processed in a period into a pair of images according to weighted average, generating a new Gait template ICGI (Improved Chrono-Gait Image), and combining a GoogleNet deep convolutional neural network to improve a space-time Gait energy map as input to extract and classify Gait recognition characteristics. The recognition effect of the gait recognition method is verified by utilizing the existing database, and the result shows that the gait recognition method is greatly improved compared with the traditional machine learning gait recognition method. The invention solves the defects of long gait recognition time based on the model, complex modeling and low cross-view angle accuracy based on the image gait recognition, and provides a method guarantee for the effective real-time identity recognition technology.
The gait recognition model is established based on the combination of the improved space-time gait energy diagram and the convolutional neural network GoogLeNet, and the structure of the gait recognition model is shown in figure 2. The model input is an improved spatio-temporal gait energy map, and because the input images may have differences in size, the images must be normalized before being subjected to learning classification. The model of the invention adopts the traditional convolutional neural network convolutional layer and pooling layer to extract features, and 2 additional fully-connected Softmax classifiers are added in the middle layer to solve the problem that the gradient disappears in the optimization process of the random gradient descent algorithm easily caused by too deep network layers. The output layer outputs the classification result by the full connection layer and the Softmax classifier. In the model training process, 2 branch classifiers added in the middle layer and the model main classifier work simultaneously, the loss function is reduced, and the network model parameters are updated. And in the process of testing the model, only using the full connection layer of the model backbone and the Softmax classifier to output a classification result.
In a further improvement, the training set of the sub-training set is set as a gait sequence of 0-180 degrees in a normal state; the gait sequence angles set by the sub-training set do not include 0 degree, 54 degrees, 90 degrees and 126 degrees; the sub-training set is set as a gait sequence of 36 degrees to 144 degrees in a normal state.
Compared with the prior method, the method has the outstanding differences and contributions that:
1. the pseudo-color coding technology is combined with the stride, the time sequence information is mapped to the lower body joint points, the original gray gait outline graph is converted into the gait outline graph with RGB colors, the color gait outline graphs in the period are evenly superposed, and the novel gait template containing the time information and the swing amplitude information between the joint points is obtained.
2. The invention relates to a method for combining an improved space-time gait energy graph with a convolutional neural network model, wherein GoogLeNet is used as a CNN network model, an inclusion module is designed as a convolutional layer, multi-scale convolution is introduced to extract multi-scale local features, and the improved space-time gait energy graph with different forms can be subjected to feature extraction and learning.
3. The experimental effects are shown in table 1: compared with the conventional gait recognition method based on gait clustering (CMCC _ 2014) and the gait recognition method based on view invariant projection (VIDP _2013, DPLCR _2018), the recognition accuracy of the cross-view gait recognition method is obviously improved. When the training set is set as 0-180 degrees gait sequence (not including test angle) in normal state, 0 degree, 54 degree, 90 degree and 126 degree, and the average accuracy rate is more than 90 percent. Compared with the gait recognition method of the deep convolutional neural network based on the gait energy map (GEI CNN _ 2017) which is popular in the recent years, the gait recognition method of the deep convolutional neural network based on the skeleton gait energy map (SGEI CNN _2018), the average accuracy is improved by more than 15%. When the training set is set with a gait sequence of 36 degrees to 144 degrees (not including the test angle) in a normal state, the average identification accuracy rate also reaches more than 95 percent, and the gait identification accuracy is improved by about 10 percent compared with the deep convolutional neural network. In addition, cross-perspective gait recognition experiments are also performed in the coat state and the knapsack state, and currently, related gait recognition experiments are lacked, so that the effectiveness of the method provided by the text is verified to be only compared with the baseline accuracy of the current CASIA-B database. The training set is set to be consistent with that in a normal state, gait sequences (not including test angles) of 36-144 degrees in the normal state are trained to be 54 degrees, 90 degrees and 126 degrees, the test set is in a coat state, 54 degrees, 90 degrees and 126 degrees in a knapsack state, and the average cross-view gait recognition accuracy in the coat state is as follows: 66.78%, the average accuracy of cross-perspective gait recognition under the backpack state is: 90.07%, compared with the baseline accuracy of the CASIA-B database, the experimental precision is obviously improved.
TABLE 1 Cross-View gait recognition under Normal State
The experimental data collection of the invention adopts a large gait database CASIA-B disclosed by the automation of Chinese academy of sciences, the CASIA-B database comprises 124 persons (93 persons for men and 31 persons for women), and each person has three walking states: normal state (containing 6 files NM1, NM2, NM3, NM4, NM5, NM6), overcoat state (containing 2 files CL1, CL2), backpack state (containing 2 files BG1, BG2), each person containing 11 perspective files (0 °, 18 °, 36 °, 54 °, 72 °, 90 °, 108 °, 126 °, 144 °, 162 °, 180 °) under each file.
Fig. 2 is a gait recognition model based on google lenet, which is implemented as follows:
the improved space-time gait energy graph is used as the input of a GoogLeNet deep convolutional neural network, a traditional CNN network basic module, namely a convolutional layer and a pooling layer, is adopted near an image input layer, the characteristic of a middle layer has a certain degree of discrimination capability, and meanwhile, the gradient disappearance problem in the optimization process of a random gradient descent algorithm caused by too deep network layers is considered, and 2 additional fully-connected Softmax classifiers are added beside a main network by the GoogLeNet. And in the model optimization process, updating network model parameters by using the sum of the loss function gradients of the trunk classifier and the branch classifiers, and in the test process, removing the corresponding branch classifiers and only using the trunk classifier to identify the gait identity.
The above embodiments are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited thereby, and any insubstantial changes and substitutions made by those skilled in the art based on the present invention are within the protection scope of the present invention.
Claims (5)
1. A gait identity recognition method based on an improved space-time gait energy diagram is characterized in that: the improved space-time gait energy diagram is combined with a convolutional neural network GoogLeNet to establish a gait recognition model, the improved space-time gait energy diagram is used as input to extract and classify gait recognition features, and the improved space-time gait energy diagram is generated by the following steps:
(1) preprocessing all gait contour maps in a gait database by image extraction, center standardization and the like;
(2) determining coordinates of a hip joint and two ankle joints of a given gait contour diagram by using an existing related algorithm, obtaining joint point positioning information, and extracting regular changes between joints of the lower half of a human body;
(3) connecting the distance between the hip joint and the left and right ankle joints to obtain the characteristics of regular change and swing amplitude between the joint points of the lower half body;
(4) combining a pseudo color coding technology with a stride, mapping time sequence information to the lower body joint points, and describing the time sequence information of a gait outline graph through pseudo colors;
(5) and (3) cycling all gait contour maps in the database through gait cycle detection, and averagely superposing the gait contour maps in the cycle after the gait contour maps are processed in the steps (1), (2) and (3) to obtain a new gait template containing time information and joint dynamic information.
2. The gait identification method based on the improved space-time gait energy map as claimed in claim 1, characterized in that: dividing the processed gait data set into a plurality of sub-test sets and sub-training sets, combining a GoogleNe deep convolutional neural network, taking a new gait template divided into the sub-training sets as input to perform gait recognition feature extraction training, and simultaneously verifying the effectiveness of the improved space-time gait energy diagram provided by the invention by using the new gait template of the sub-test sets.
3. A gait identification method based on improved space-time gait energy diagram according to claim 2, characterized in that: and the sub-training set is set as a 0-180-degree gait sequence in a normal state.
4. A gait identification method based on improved space-time gait energy diagram according to claim 3, characterized in that: the gait sequence angles set by the sub-training set do not include 0 degrees, 54 degrees, 90 degrees and 126 degrees.
5. The gait identification method based on the improved space-time gait energy map as claimed in claim 4, characterized in that: the sub-training set is set as a gait sequence of 36 degrees to 144 degrees in a normal state.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010994932.8A CN112163487A (en) | 2020-09-21 | 2020-09-21 | Gait identity recognition method based on improved space-time gait energy diagram |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010994932.8A CN112163487A (en) | 2020-09-21 | 2020-09-21 | Gait identity recognition method based on improved space-time gait energy diagram |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112163487A true CN112163487A (en) | 2021-01-01 |
Family
ID=73863100
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010994932.8A Pending CN112163487A (en) | 2020-09-21 | 2020-09-21 | Gait identity recognition method based on improved space-time gait energy diagram |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112163487A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113111797A (en) * | 2021-04-19 | 2021-07-13 | 杭州电子科技大学 | Cross-view gait recognition method combining self-encoder and view transformation model |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104463118A (en) * | 2014-12-04 | 2015-03-25 | 龙岩学院 | Multi-view-angle gait recognition method based on Kinect |
| WO2017006314A1 (en) * | 2015-07-05 | 2017-01-12 | THE WHOLLYSEE.Ltd. | Optical identification and characterization system and tagss |
| CN108564005A (en) * | 2018-03-26 | 2018-09-21 | 电子科技大学 | A kind of human body tumble discrimination method based on convolutional neural networks |
| US20190172232A1 (en) * | 2017-12-05 | 2019-06-06 | Samsung Eletrônica da Amazônia Ltda. | Method and system for sensor data recognition using data enrichment for the learning process |
| CN110097029A (en) * | 2019-05-14 | 2019-08-06 | 西安电子科技大学 | Identity identifying method based on Highway network multi-angle of view Gait Recognition |
| CN111259700A (en) * | 2018-12-03 | 2020-06-09 | 北京京东尚科信息技术有限公司 | Method and apparatus for generating gait recognition model |
-
2020
- 2020-09-21 CN CN202010994932.8A patent/CN112163487A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104463118A (en) * | 2014-12-04 | 2015-03-25 | 龙岩学院 | Multi-view-angle gait recognition method based on Kinect |
| WO2017006314A1 (en) * | 2015-07-05 | 2017-01-12 | THE WHOLLYSEE.Ltd. | Optical identification and characterization system and tagss |
| US20190172232A1 (en) * | 2017-12-05 | 2019-06-06 | Samsung Eletrônica da Amazônia Ltda. | Method and system for sensor data recognition using data enrichment for the learning process |
| CN108564005A (en) * | 2018-03-26 | 2018-09-21 | 电子科技大学 | A kind of human body tumble discrimination method based on convolutional neural networks |
| CN111259700A (en) * | 2018-12-03 | 2020-06-09 | 北京京东尚科信息技术有限公司 | Method and apparatus for generating gait recognition model |
| CN110097029A (en) * | 2019-05-14 | 2019-08-06 | 西安电子科技大学 | Identity identifying method based on Highway network multi-angle of view Gait Recognition |
Non-Patent Citations (1)
| Title |
|---|
| 蒋敏兰等: "基于改进时空步态图的步态识别", 《信息技术与网络安全》, vol. 38, no. 12, pages 53 - 55 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113111797A (en) * | 2021-04-19 | 2021-07-13 | 杭州电子科技大学 | Cross-view gait recognition method combining self-encoder and view transformation model |
| CN113111797B (en) * | 2021-04-19 | 2024-02-13 | 杭州电子科技大学 | Cross-view gait recognition method combining self-encoder and view transformation model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108921100B (en) | Face recognition method and system based on visible light image and infrared image fusion | |
| CN110135375B (en) | Multi-person attitude estimation method based on global information integration | |
| CN108520216B (en) | Gait image-based identity recognition method | |
| CN107194341B (en) | Face recognition method and system based on fusion of Maxout multi-convolution neural network | |
| CN105574518B (en) | Method and device for detecting living human face | |
| CN108256421A (en) | A kind of dynamic gesture sequence real-time identification method, system and device | |
| CN110084156A (en) | A kind of gait feature abstracting method and pedestrian's personal identification method based on gait feature | |
| CN110516571A (en) | Inter-library micro- expression recognition method and device based on light stream attention neural network | |
| CN108182397B (en) | Multi-pose multi-scale human face verification method | |
| CN105138954A (en) | Image automatic screening, query and identification system | |
| CN114187665B (en) | Multi-person gait recognition method based on human skeleton heat map | |
| CN106056053A (en) | Human posture recognition method based on skeleton feature point extraction | |
| CN110008913A (en) | The pedestrian's recognition methods again merged based on Attitude estimation with viewpoint mechanism | |
| CN111126240B (en) | Three-channel feature fusion face recognition method | |
| CN114067358A (en) | Human body posture recognition method and system based on key point detection technology | |
| CN114998934B (en) | Clothes-changing pedestrian re-identification and retrieval method based on multi-mode intelligent perception and fusion | |
| CN111191564A (en) | Multi-pose face emotion recognition method and system based on multi-angle neural network | |
| CN110796101A (en) | Face recognition method and system of embedded platform | |
| CN113920326A (en) | Tumble behavior identification method based on human skeleton key point detection | |
| CN108960076A (en) | Ear recognition and tracking based on convolutional neural networks | |
| CN110263605A (en) | Pedestrian's dress ornament color identification method and device based on two-dimension human body guise estimation | |
| CN105069745A (en) | face-changing system based on common image sensor and enhanced augmented reality technology and method | |
| CN106815855A (en) | Based on the human body motion tracking method that production and discriminate combine | |
| CN109325408A (en) | A kind of gesture judging method and storage medium | |
| CN112446302A (en) | Human body posture detection method and system, electronic equipment and storage medium |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210101 |
|
| WD01 | Invention patent application deemed withdrawn after publication |