CN109815815A - A kind of pedestrian being integrated based on metric learning and support vector machines recognition methods again - Google Patents
A kind of pedestrian being integrated based on metric learning and support vector machines recognition methods again Download PDFInfo
- Publication number
- CN109815815A CN109815815A CN201811576219.0A CN201811576219A CN109815815A CN 109815815 A CN109815815 A CN 109815815A CN 201811576219 A CN201811576219 A CN 201811576219A CN 109815815 A CN109815815 A CN 109815815A
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- support vector
- vector machines
- metric learning
- linear space
- 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
- 238000012706 support-vector machine Methods 0.000 title claims abstract description 52
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000012545 processing Methods 0.000 claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims abstract description 9
- 238000005259 measurement Methods 0.000 claims abstract description 5
- 230000000007 visual effect Effects 0.000 claims description 23
- 230000000694 effects Effects 0.000 claims description 14
- 238000012549 training Methods 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 abstract description 7
- 102100040160 Rabankyrin-5 Human genes 0.000 description 12
- 101710086049 Rabankyrin-5 Proteins 0.000 description 12
- 241000271897 Viperidae Species 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- SYJGKVOENHZYMQ-UHFFFAOYSA-N Penoxsulam Chemical compound N1=C2C(OC)=CN=C(OC)N2N=C1NS(=O)(=O)C1=C(OCC(F)F)C=CC=C1C(F)(F)F SYJGKVOENHZYMQ-UHFFFAOYSA-N 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- MXCVHSXCXPHOLP-UHFFFAOYSA-N 4-oxo-6-propylchromene-2-carboxylic acid Chemical compound O1C(C(O)=O)=CC(=O)C2=CC(CCC)=CC=C21 MXCVHSXCXPHOLP-UHFFFAOYSA-N 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 101100478297 Arabidopsis thaliana SR45 gene Proteins 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000004438 eyesight Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of pedestrian being integrated based on metric learning and support vector machines recognition methods again, belong to image procossing, mode identification technology.The present invention firstly generates pedestrian's eigenmatrix with pedestrian's label information；The non-linear space M of measurement pedestrian's distance is processed；The pedestrian's label information used in support vector machines is set；Support vector machines introduces bound variable, then using support vector machines as the constraint condition of non-linear space；Processing is zoomed in and out to the constraint condition of non-linear space M；The optimal solution for finding projection matrix and classifier carries out pedestrian's identification with the identification model that metric learning and support vector machines are integrated, obtains discrimination.The present invention is integrated with metric learning and support vector machines.It is compared with existing method, the label information that method proposed by the invention is effectively excavated, is utilized in pedestrian's data set promotes pedestrian's matching rate effectively.
Description
Technical field
The present invention relates to a kind of pedestrian being integrated based on metric learning and support vector machines recognition methods again, belong to image
Processing, mode identification technology.
Background technique
With the popularization of national smart city, safe city, video monitoring system covers China main cities substantially.Video
The pedestrian information data that monitoring system preserves are huge, and artificial treatment information efficiency low cost is big, and identification technology can again by pedestrian
Effectively to improve working efficiency, save resource.Pedestrian identifies that main task is the row matched under non-overlap video camera again
Whether people is the same pedestrian.The technology is to extract a pedestrian out under a video camera as target pedestrian, is judged in b video camera
Under whether there is the pedestrian, if there is then finding target pedestrian.Therefore, pedestrian again identification technology by vast research
The very big concern of person.
Although currently, identify the attention for causing researcher again with the pedestrian of higher robustness, and propose it is some can
Capable solution, but experiment effect is not able to satisfy the needs of reality still, especially when pedestrian deliberately goes to change oneself external spy
When sign performance.Identification technology still has variation due to visual angle and illumination and camera parameter setting to existing pedestrian again
Difference causes same a group traveling together to show different bottom views under different visions, different illumination conditions and different cameras
Feel feature, shows similar visual signature without same pedestrian, identification technology is far from satisfying reality again so as to cause pedestrian
The needs of application.
Study of recognition achievement mainly has that the pedestrian based on feature identifies again and the pedestrian based on metric learning knows again to pedestrian again
Other two major classes in identification are to extract discriminating power and ability to express all more from pedestrian's low-level image feature based on the pedestrian of feature
Strong feature.It is from the better algorithm of metric learning angle set off in search matching effect that pedestrian based on metric learning identifies again.
Recognition methods utilizes the label information of pedestrian to extract pedestrian's low-level image feature to pedestrian based on feature again, but it is violent to be illuminated by the light variation
The influence of the factors such as different is set with camera parameters, and the low-level image features such as the color of pedestrian and texture change very greatly, lead to pedestrian
Recognition accuracy is low.However, the pedestrian based on metric learning again recognition methods in view of illumination variation acutely, camera parameters set
The problems such as setting different and pedestrian's appearance is reduced same using the label information occurred in pairs between pedestrian same under different perspectives
Metric range between a group traveling together.But under only considered different perspectives when using pedestrian's label information the same pedestrian mark
Information is signed, there is no the label informations considered under different perspectives between different pedestrians.Know again for the pedestrian based on metric learning
Other algorithm is under the disturbing factor that pedestrian wears the complex backgrounds such as dress ornament, posture be similar clothes, and recognizer recognition effect is poor again by pedestrian,
The not strong problem of robustness.
Summary of the invention
The present invention provides a kind of pedestrian being integrated based on metric learning and support vector machines recognition methods again, excavate simultaneously
The label information under different perspectives between pedestrian is made full use of to improve the accuracy rate of pedestrian's identification.
The technical scheme is that a kind of pedestrian being integrated based on metric learning and support vector machines side of identification again
Method firstly generates pedestrian's eigenmatrix with pedestrian's label information；The non-linear space M of measurement pedestrian's distance is processed；
The pedestrian's label information used in support vector machines is set；Support vector machines introduce bound variable, then using support vector machines as
The constraint condition of non-linear space；Processing is zoomed in and out to the constraint condition of non-linear space M；Find projection matrix and classifier
Optimal solution, carry out pedestrian's identification with the identification model that metric learning and support vector machines are integrated, obtain discrimination.
Further, the specific steps of the pedestrian being integrated based on metric learning and support vector machines recognition methods again
It is as follows:
Stpe1, a, the feature of pedestrian all projects to the same non-linear space M ∈ R under the visual angle bm×nIt is interior, utilize public affairs
FormulaFind out under the visual angle a with it is most like under the visual angle b but be not oneself pedestrianTo generate band
There is pedestrian's eigenmatrix x of pedestrian's label informationc；
Stpe2, in non-linear space M, it is desirable that i-th of people under each visual angle aWith under the visual angle b withMost
Pedestrian that is similar but not being the same personBetween metric range be less than i-th of people under the visual angle aExist with own
Under the visual angle bBetween metric range, i.e.,
If Stpe3, all pedestrian's features all press Stpe2 processing, the feature after projection inherently meets above situation
Just will appear over-fitting situation, thus if projection after feature inherently meet the same pedestrian wear difference it is huge if by
Step2 processing, it is without any processing if different pedestrian's dresses are similar,Take 0；I.e.
Stpe4, pedestrian label information set-up mode are as follows: ifWithIt is same people, then their label is set
Information yijBe 1, if not then be -1；I.e.
Wherein,It indicates in support vector machinesMetric range between pedestrian's feature；
Stpe5, traditional support vector machine introduce a bound variable ξijAs inequalityWherein, w is the classifier of support vector machines；The constraint condition of non-linear space M is drawn again
Enter in support vector machines, it may be assumed that
s.t(yij(w(Mxai-Mxbj)+c) > 1- ξij)
Stpe6, the constraint condition (y non-linear space Mij(w(Mxai-Mxbj)+c) > 1- ξij) carry out scaling appropriate
Processing, by ξ after constraint relaxation is optimizedijIt eliminates, i.e.,
Stpe7, the optimal solution for finding Stpe6 formula, training study obtain projection matrix and classifier w；Again with projection square
Battle array and classifier w carry out pedestrian's identification in the identification model that metric learning and support vector machines are integrated, and obtain discrimination s,
The identification model that the metric learning and support vector machines for wherein obtaining discrimination are integrated is as follows:
Wherein, the effect of c is the range of limitation classification similar value.
The beneficial effects of the present invention are:
The present invention is integrated with metric learning and support vector machines.It is compared with existing method, method proposed by the invention has
The excavation of effect, the label information being utilized in pedestrian's data set, promote pedestrian's matching rate effectively.
Detailed description of the invention
Fig. 1 is flow diagram of the present invention；Wherein, y is the pedestrian's label information matrix for indicating to use inside support vector machines；
The model that Fig. 2 is metric learning model and metric learning and support vector machines is integrated is respectively on four data sets
Contrast and experiment.
Specific embodiment
Embodiment 1: as shown in Figs. 1-2, a kind of pedestrian being integrated based on the metric learning and support vector machines side of identification again
Method firstly generates pedestrian's eigenmatrix with pedestrian's label information；The non-linear space M of measurement pedestrian's distance is processed；
The pedestrian's label information used in support vector machines is set；Support vector machines introduce bound variable, then using support vector machines as
The constraint condition of non-linear space；Processing is zoomed in and out to the constraint condition of non-linear space M；Find projection matrix and classifier
Optimal solution, carry out pedestrian's identification with the identification model that metric learning and support vector machines are integrated, obtain discrimination.
Further, the specific steps of the pedestrian being integrated based on metric learning and support vector machines recognition methods again
It is as follows:
Stpe1, a, the feature of pedestrian all projects to the same non-linear space M ∈ R under the visual angle bm×nIt is interior, utilize public affairs
FormulaFind out under the visual angle a with it is most like under the visual angle b but be not oneself pedestrianTo generate band
There is pedestrian's eigenmatrix x of pedestrian's label informationc；
Stpe2, in non-linear space M, it is desirable that i-th of people under each visual angle aWith under the visual angle b withMost
Pedestrian that is similar but not being the same personBetween metric range be less than i-th of people under the visual angle a
With own under the visual angle bBetween metric range, i.e.,
If Stpe3, all pedestrian's features all press Stpe2 processing, the feature after projection inherently meets above situation
Just will appear over-fitting situation, thus if projection after feature inherently meet the same pedestrian wear difference it is huge if by
Step2 processing, it is without any processing if different pedestrian's dresses are similar,Take 0；I.e.
Stpe4, pedestrian label information set-up mode are as follows: ifWithIt is same people, then their label is set
Information yijBe 1, if not then be -1；I.e.
Wherein,It indicates in support vector machinesMetric range between pedestrian's feature；
Stpe5, traditional support vector machine introduce a bound variable ξijAs inequalityWherein, w is the classifier of support vector machines；The constraint condition of non-linear space M is drawn again
Enter in support vector machines, it may be assumed that
s.t(yij(w(Mxai-Mxbj)+c) > 1- ξij)
Stpe6, the constraint condition (y non-linear space Mij(w(Mxai-Mxbj)+c) > 1- ξij) carry out scaling appropriate
Processing, by ξ after constraint relaxation is optimizedijIt eliminates, i.e.,
Stpe7, the optimal solution for finding Stpe6 formula, training study obtain projection matrix and classifier w；Again with projection square
Battle array and classifier w carry out pedestrian's identification in the identification model that metric learning and support vector machines are integrated, and obtain discrimination s,
The identification model that the metric learning and support vector machines for wherein obtaining discrimination are integrated is as follows:
Wherein, the effect of c is the range of limitation classification similar value.
In order to be compared with existing method, the present invention using VIPER, iLIDS-IVD, CUHK01, PRID2011 and
The enterprising every trade people of five data sets of PRID405S identifies experiment again, and is used as using the average value of ten folding cross validations and is most terminated
Fruit.
Evaluation index is consistent with control methods, using CMC summation curve as evaluation index (although CMC curve graph is
The matching rate of rank1-rank20 is listed, but in actual picture retrieval etc. is used, that often most value is rank1
Value).
In order to verify the effect of support vector machines, using the measurement model of proposition as the (Ours- shown in Fig. 2 of model one
Svm), metric learning adds the model of support vector machines as model two (Ours shown in Fig. 2).Two models VIPER,
Shown in CMC curve graph 2 on this four data sets of iLIDS-IVD, PRID2011 and PRID405S.It can be seen that being spent before rank5
Amount study plus support vector machines are obvious to pedestrian's discrimination effect is improved, as the increase of rank value gradually tends towards stability.
On VIPeR data set, PCCA, LFDA, KISSME, LADF, Mid-filter, ECM, MFA, kLFDA, RD and SR45
Model compares, and the results are shown in Table 1.Although CMC curve graph is to list the matching rate of rank1-rank20, in practical figure
Piece retrieval etc. use in, what is often most valued is the value of rank1, as can be seen from Table 1 it is proposed that method rank1 and
Rank5 effect is best.
Table 1: comparison result of the various methods on VIPER data set, list rank1, rank5, rank10 and
The matching rate (%) of rank20
| Methods | Rank1 | Rank5 | Rank10 | Rank20 |
| PCCA | 19.3 | 48.9 | 64.9 | 80.3 |
| LFDA | 19.7 | 46.7 | 62.1 | 77.0 |
| KISSME | 19.6 | 48.0 | 62.2 | 77.0 |
| LADF | 29.3 | 61.0 | 76.0 | 86.2 |
| Mid-filter | 29.1 | 52.3 | 66.0 | 79.9 |
| ECM | 38.2 | 67.2 | 78.3 | 87.9 |
| MFA | 32.2 | 66.0 | 79.7 | 90.6 |
| RD | 33.3 | 41.5 | 78.4 | 88.5 |
| kLFDA | 32.3 | 65.8 | 79.7 | 90.6 |
| SR | 32.9 | 62.0 | 75.9 | 89.2 |
| Ours | 55.5 | 67.2 | 73.2 | 80.5 |
On iLIDS-IDV data set, SDALF-SS, Color+LBP+DTW, ISR, DVR, DVDL, PHDL+WHOS+
STFV3D, Salience+DVR, KISSME compare, and the results are shown in Table 2.Method proposed by the present invention as can be seen from Table 2
Rank1 and rank5 discrimination is all highest.
Table 2: comparison result of the various methods on iLIDS-IVD data set, list rank1, rank5, rank10 and
The matching rate (%) of rank20
On CUHK01 data set experiment with Rcca, ITML, KISSME, GenericMetric68, SalMatch, kLFDA,
MidFilter, MirrorKMFA, LOMO+LADF, LOMO+XQDA comparison result are as shown in table 3, as can be seen from Table 3 this hair
The method of bright proposition recognition effect on rank1 is best.
Table 3: comparison result of the various methods on CUHK01 data set, list rank1, rank5, rank10 and
The matching rate (%) of rank20
| Methods | Rank1 | Rank5 | Rank10 | Rank20 |
| Rcca | 14.9 | 32.6 | 43.8 | 55.5 |
| ITML | 16.0 | 35.2 | 45.6 | 59.8 |
| KISSME | 10.3 | 27.2 | 37.5 | 49.7 |
| GenericMetric | 20.0 | 43.6 | 56.0 | 69.3 |
| SalMatch | 28.5 | 45.9 | 55.7 | 68.0 |
| kLFDA | 26.1 | 49.4 | 58.4 | 71.8 |
| MidFilter | 34.3 | 55.1 | 65.0 | 74.9 |
| MirrorKMFA | 40.4 | 64.6 | 75.3 | 84.1 |
| LOMO+LADF | 58.0 | 83.7 | 90.5 | 94.9 |
| LOMO+XQDA | 63.2 | 83.9 | 90.0 | 94.4 |
| Ours | 64.5 | 81.9 | 85 | 88 |
On 2011 data set of PRID, PPLM, RDC, LOMO+LADF, MetricEnsemble, LOMO+M, XQDA, LOMO+
XQDA, DVR, Salience+DVR compare, and the results are shown in Table 4, in addition to rank20 effect is not satisfactory, the method for the present invention
Rank1, rank2 and rank10 recognition effect are all best.
Table 4: comparison result of the various methods on 2011 data set of PRID, list rank1, rank5, rank10 and
The matching rate (%) of rank20
| Methods | Rank1 | Rank5 | Rank10 | Rank20 |
| PPLM | 15.0 | 32.0 | 42.0 | 54.0 |
| RDC | 15.5 | 38.8 | 53.2 | 69.0 |
| LOMO+LADF | 16.2 | 34.0 | 44.4 | 59.5 |
| MetricEnsemble | 17.9 | 39.0 | 50.0 | 62.0 |
| LOMO+M | 15.2 | 36.1 | 48.3 | 60.4 |
| XQDA | 24.6 | 49.3 | 62.8 | 76.3 |
| LOMO+XQDA | 26.7 | 49.9 | 61.9 | 73.8 |
| DVR | 28.9 | 55.3 | 65.5 | 82.8 |
| Salience+DVR | 41.7 | 64.5 | 77.5 | 88.8 |
| Ours | 72.3 | 86.8 | 92 | 96.7 |
On PRID450S data set, ELF, KISSME, EIML, SCNCD, ECM, TSR, MEDVL, KISSME-MGT, LOMO+
The methods of LADF, KLFDA-MGT, MirrorKMFA compare, and the results are shown in Table 5, and the method for the present invention rank1 effect is best.
Table 5: comparison result of the various methods on PRID_450S data set, list rank1, rank5, rank10 and
The matching rate (%) of rank20
| Methods | Rank1 | Rank5 | Rank10 | Rank20 |
| ELF | 30.6 | - | 73.6 | 84.2 |
| KISSME | 33.0 | - | 71.0 | 79.0 |
| EIML | 35.0 | - | 68.0 | 77 |
| SCNCD | 41.5 | 66.6 | 75.9 | 84.4 |
| ECM | 41.9 | 66.3 | 76.9 | 84.9 |
| TSR | 44.9 | 71.7 | 77.5 | 86.7 |
| MEDVL | 45.9 | 73.0 | 82.9 | 91.1 |
| KISSME-MGT | 46.1 | 73.3 | 83.3 | 90.7 |
| LOMO+LADF | 47.8 | 74.7 | 82.8 | 90.9 |
| KLFDA-MGT | 46.1 | 73.3 | 83.3 | 90.7 |
| MirrorKMFA | 55.4 | 79.3 | 87.8 | 93.9 |
| Ours | 58.4 | 68.7 | 73.1 | 80.2 |
Above in conjunction with attached drawing, the embodiment of the present invention is explained in detail, but the present invention is not limited to above-mentioned
Embodiment within the knowledge of a person skilled in the art can also be before not departing from present inventive concept
Put that various changes can be made.
Claims (2)
1. a kind of pedestrian being integrated based on metric learning and support vector machines recognition methods again, it is characterised in that: firstly generate
Pedestrian's eigenmatrix with pedestrian's label information；The non-linear space M of measurement pedestrian's distance is processed；Supporting vector is set
The pedestrian's label information used in machine；Support vector machines introduces bound variable, then using support vector machines as non-linear space
Constraint condition；Processing is zoomed in and out to the constraint condition of non-linear space M；The optimal solution of projection matrix and classifier is found, is used
The identification model that metric learning and support vector machines are integrated carries out pedestrian's identification, obtains discrimination.
2. the pedestrian being integrated based on metric learning and support vector machines recognition methods again according to claim 1, special
Sign is: specific step is as follows for recognition methods again by the pedestrian being integrated based on metric learning and support vector machines:
Stpe1, a, the feature of pedestrian all projects to the same non-linear space M ∈ R under the visual angle bm×nIt is interior, utilize formulaFind out under the visual angle a with it is most like under the visual angle b but be not oneself pedestrianIt is had to generate
Pedestrian's eigenmatrix x of pedestrian's label informationc；
Stpe2, in non-linear space M, it is desirable that i-th of people under each visual angle aWith under the visual angle b withMost like
It but is not the pedestrian of the same personBetween metric range be less than i-th of people under the visual angle aWith own at the visual angle b
UnderBetween metric range, i.e.,
If Stpe3, all pedestrian's features all press Stpe2 processing, the feature after projection inherently meets above situation will
Occur over-fitting situation, so if projection after feature inherently meet the same pedestrian wear difference it is huge if by Step2
Processing, it is without any processing if different pedestrian's dresses are similar,
Take 0；I.e.
Stpe4, pedestrian label information set-up mode are as follows: ifWithIt is same people, then their label information is set
yijBe 1, if not then be -1；I.e.
Wherein,It indicates in support vector machinesMetric range between pedestrian's feature；
Stpe5, traditional support vector machine introduce a bound variable ξijAs inequalityIts
In, w is the classifier of support vector machines；The constraint condition of non-linear space M is introduced into support vector machines again, it may be assumed that
s.t(yij(w(Mxai-Mxbj)+c) > 1- ξij)
Stpe6, the constraint condition (y non-linear space Mij(w(Mxai-Mxbj)+c) > 1- ξij) scaling processing appropriate is carried out,
By ξ after constraint relaxation is optimizedijIt eliminates, i.e.,
Stpe7, the optimal solution for finding Stpe6 formula, training study obtain projection matrix and classifier w；Again with projection matrix and
Classifier w carries out pedestrian's identification in the identification model that metric learning and support vector machines are integrated, and obtains discrimination s, wherein
The identification model that the metric learning and support vector machines for obtaining discrimination are integrated is as follows:
Wherein, the effect of c is the range of limitation classification similar value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811576219.0A CN109815815B (en) | 2018-12-22 | 2018-12-22 | Pedestrian re-identification method based on metric learning and support vector machine integration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811576219.0A CN109815815B (en) | 2018-12-22 | 2018-12-22 | Pedestrian re-identification method based on metric learning and support vector machine integration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109815815A true CN109815815A (en) | 2019-05-28 |
| CN109815815B CN109815815B (en) | 2021-06-18 |
Family
ID=66602380
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811576219.0A Active CN109815815B (en) | 2018-12-22 | 2018-12-22 | Pedestrian re-identification method based on metric learning and support vector machine integration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109815815B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080101705A1 (en) * | 2006-10-31 | 2008-05-01 | Motorola, Inc. | System for pattern recognition with q-metrics |
| US20120155719A1 (en) * | 2010-12-21 | 2012-06-21 | Electronics And Telecommunications Research Institute | Apparatus and method for detecting upper body |
| CN106778921A (en) * | 2017-02-15 | 2017-05-31 | 张烜 | Personnel based on deep learning encoding model recognition methods again |
| WO2017212206A1 (en) * | 2016-06-06 | 2017-12-14 | Cirrus Logic International Semiconductor Limited | Voice user interface |
| CN107844752A (en) * | 2017-10-20 | 2018-03-27 | 常州大学 | A kind of recognition methods again of the pedestrian based on block rarefaction representation |
| US20180129648A1 (en) * | 2016-09-12 | 2018-05-10 | Sriram Chakravarthy | Methods and systems of automated assistant implementation and management |
| US20180173940A1 (en) * | 2016-12-19 | 2018-06-21 | Canon Kabushiki Kaisha | System and method for matching an object in captured images |
| CN108345860A (en) * | 2018-02-24 | 2018-07-31 | 江苏测联空间大数据应用研究中心有限公司 | Personnel based on deep learning and learning distance metric recognition methods again |
| CN108509854A (en) * | 2018-03-05 | 2018-09-07 | 昆明理工大学 | A kind of constrained based on projection matrix combines the pedestrian's recognition methods again for differentiating dictionary learning |
-
2018
- 2018-12-22 CN CN201811576219.0A patent/CN109815815B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080101705A1 (en) * | 2006-10-31 | 2008-05-01 | Motorola, Inc. | System for pattern recognition with q-metrics |
| US20120155719A1 (en) * | 2010-12-21 | 2012-06-21 | Electronics And Telecommunications Research Institute | Apparatus and method for detecting upper body |
| WO2017212206A1 (en) * | 2016-06-06 | 2017-12-14 | Cirrus Logic International Semiconductor Limited | Voice user interface |
| US20180129648A1 (en) * | 2016-09-12 | 2018-05-10 | Sriram Chakravarthy | Methods and systems of automated assistant implementation and management |
| US20180173940A1 (en) * | 2016-12-19 | 2018-06-21 | Canon Kabushiki Kaisha | System and method for matching an object in captured images |
| CN106778921A (en) * | 2017-02-15 | 2017-05-31 | 张烜 | Personnel based on deep learning encoding model recognition methods again |
| CN107844752A (en) * | 2017-10-20 | 2018-03-27 | 常州大学 | A kind of recognition methods again of the pedestrian based on block rarefaction representation |
| CN108345860A (en) * | 2018-02-24 | 2018-07-31 | 江苏测联空间大数据应用研究中心有限公司 | Personnel based on deep learning and learning distance metric recognition methods again |
| CN108509854A (en) * | 2018-03-05 | 2018-09-07 | 昆明理工大学 | A kind of constrained based on projection matrix combines the pedestrian's recognition methods again for differentiating dictionary learning |
Non-Patent Citations (2)
| Title |
|---|
| DAPENG TAO 等: "Person Re-Identification by Dual-Regularized KISS Metric Learning", 《IEEE TRANSACTIONS ON IMAGE PROCESSING》 * |
| 王宇凡: "基于相似度量的模糊支持向量机算法研究", 《微电子学与计算机》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109815815B (en) | 2021-06-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Raja et al. | Color object detection based image retrieval using ROI segmentation with multi-feature method | |
| Kumar et al. | Face detection techniques: a review | |
| Chen et al. | Saliency detection via the improved hierarchical principal component analysis method | |
| Pflug et al. | Ear biometrics: a survey of detection, feature extraction and recognition methods | |
| Varish et al. | Image retrieval scheme using quantized bins of color image components and adaptive tetrolet transform | |
| CN105718889B (en) | Based on GB (2D)2The face personal identification method of PCANet depth convolution model | |
| Phimoltares et al. | Face detection and facial feature localization without considering the appearance of image context | |
| Tian et al. | Natural scene text detection with MC–MR candidate extraction and coarse-to-fine filtering | |
| CN106022375A (en) | HU invariant moment and support vector machine-based garment style identification method | |
| CN101694691A (en) | Method and device for synthesizing facial images | |
| CN102521616A (en) | Pedestrian detection method on basis of sparse representation | |
| CN110046544A (en) | Digital gesture identification method based on convolutional neural networks | |
| CN108052912A (en) | A kind of three-dimensional face image recognition methods based on square Fourier descriptor | |
| CN106056132A (en) | Garment style identification method based on Fourier descriptor and support vector machine | |
| Lahiani et al. | Hand pose estimation system based on Viola-Jones algorithm for android devices | |
| Lee et al. | Property-specific aesthetic assessment with unsupervised aesthetic property discovery | |
| Murphy-Chutorian et al. | N-tree Disjoint-Set Forests for Maximally Stable Extremal Regions. | |
| Fang et al. | Pedestrian attributes recognition in surveillance scenarios with hierarchical multi-task CNN models | |
| CN110633631B (en) | Pedestrian re-identification method based on component power set and multi-scale features | |
| Zheng et al. | Weight-based sparse coding for multi-shot person re-identification | |
| CN108564020B (en) | Micro-gesture recognition method based on panoramic 3D image | |
| CN109815815A (en) | A kind of pedestrian being integrated based on metric learning and support vector machines recognition methods again | |
| CN101976339B (en) | Local characteristic extraction method for face recognition | |
| Molina-Giraldo et al. | Video segmentation based on multi-kernel learning and feature relevance analysis for object classification | |
| Kobyshev et al. | 3D saliency for finding landmark buildings |
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 |