CN104881655A - Human behavior recognition method based on multi-feature time-space relationship fusion - Google Patents
Human behavior recognition method based on multi-feature time-space relationship fusion Download PDFInfo
- Publication number
- CN104881655A CN104881655A CN201510298003.2A CN201510298003A CN104881655A CN 104881655 A CN104881655 A CN 104881655A CN 201510298003 A CN201510298003 A CN 201510298003A CN 104881655 A CN104881655 A CN 104881655A
- Authority
- CN
- China
- Prior art keywords
- time
- space
- barycenter
- features
- space relationship
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
- G06V20/42—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items of sport video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/49—Segmenting video sequences, i.e. computational techniques such as parsing or cutting the sequence, low-level clustering or determining units such as shots or scenes
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Computing Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a human behavior recognition method based on multi-feature time-space relationship fusion, specifically comprising the following steps: first, dense trajectory features extracted from videos are represented through an optical flow histogram and a moving boundary histogram; then, a time-space bipartite graph between the centers of mass of the two types of features is constructed by a KMEANS algorithm, the time-space bipartite graph is segmented by a K-path bipartite graph segmentation technology, and a representation method based on conditional probability is adopted to obtain video-level coding after fusion of the two types of features; and finally, a classifier is trained, and recognition is performed. According to the human behavior recognition method based on multi-feature time-space relationship fusion, the time-space bipartite graph between the centers of mass of the two types of features is constructed by calculating the time-space distance between the features of each video, the time-space bipartite graph is segmented by the K-path bipartite graph segmentation technology, and the centers of mass with strong time-space relationship are fused. Thus, effective information of different features is better mined, and the recognition accuracy is improved.
Description
Technical field
The present invention relates to computer vision field, especially relate to a kind of human behavior recognition methods of merging based on multiple features time-space relationship.
Background technology
Along with the development of computer science, video starts to become a part for people's life, how to allow human behavior in computing machine " understanding " video, all has vital role to fields such as content based video retrieval system, intelligent monitoring, man-machine interaction and virtual realities.
Generally speaking, a classical human behavior identification framework mainly comprises three steps: the training of feature extraction, Video coding and sorter and identification, in addition, for the manifold situation of employing, also comprise an optional multiple features to merge or later stage fusion steps early stage, Video coding is wherein the committed step determining recognition accuracy.
At present, one of coding method being widely used and improving is word bag (BagofWords, be called for short BoW) method, first classical BoW method carries out cluster to feature, then videometer be shown as feature appear in each barycenter frequency histogram vector, although BoW coding shows good generalization ability and robustness in a lot of document, but the method also has a lot of shortcoming: such as time-consuming feature clustering process, the loss having time-space relationship information between supervision parameter k and barycenter of KMEANS algorithm.
In order to the problem that the parameter k dependence experience eliminating KMEANS algorithm is determined, " LiuJ, ShahM.Learninghumanactionsviainformationmaximization [C] .ComputerVisionandPatternRecognition, 2008.CVPR2008.IEEEConferenceon.IEEE, 2008:1-8. " use that mutual information maximization clustering algorithm is unsupervised determines most suitable barycenter quantity, first this algorithm carries out KMEANS cluster with a larger k, to reduce the information dropout because KMEANS cluster causes, under the prerequisite of the least possible drop-out, barycenter quantity is reduced afterwards by mutual information maximization clustering algorithm, the computing velocity of subsequent step is improved with this.
In order to solve the problem of time-space relationship information dropout, many researchers propose the extended method based on BoW, according to the difference of retained information, these methods are divided into two classes: the BoW retaining absolute space-time information represents and the BoW that retains relative space-time information represents.The former needs to carry out global segmentation to the space-time body of video usually, and this makes the absolute space-time coordinate of Video coding and the feature calculated relevant, lacks translation invariance." LaptevI; MarszalekM; SchmidC; etal.Learningrealistichumanactionsfrommovies [C] .ComputerVisionandPatternRecognition; 2008.CVPR2008.IEEEConferenceon.IEEE; 2008:1-8. " is slit into predefined space-time grid the space-time body integration of video, calculates BoW respectively afterwards and the BoW of all grids vector is together in series as final Video coding in each grid.But in order to determine best grid combination, the method needs to carry out greed search with cross validation, and this step is very time-consuming, and in addition, the overlength vector obtained by the BoW of different grid of connecting further increases computation complexity." SunJ; WuX; YanS; etal.Hierarchicalspatio-temporalcontextmodelingforaction recognition [C] .ComputerVisionandPatternRecognition; 2009.CVPR2009.IEEEConferenceon.IEEE, 2009:2004-2011. " obtains three layers of spatio-temporal context information in a kind of mode of level.And the latter, namely retain the method for relative space-time information, normally utilize the relative space-time distance between BoW barycenter or feature to carry out Video coding." KovashkaA; GraumanK.Learningahierarchyofdiscriminativespace-timenei ghborhoodfeaturesforhumanactionrecognition [C] .ComputerVisionandPatternRecognition (CVPR); 2010IEEEConferenceon.IEEE; 2010:2046-2053. " first with the some structure new feature around former unique point, afterwards in conjunction with barycenter information belonging to new feature and new feature directional information structure Video coding.Owing to needing the barycenter constructing multi-layer, the computation complexity of the method is relatively high." WangJ; ChenZ; WuY.Actionrecognitionwithmultiscalespatio-temporalcontex ts [C] .ComputerVisionandPatternRecognition (CVPR); 2011IEEEConferenceon.IEEE, 2011:3185-3192. " carries out Video coding by the space-time context interactive information obtained in multiple spatial and temporal scales of former feature between feature.。
Summary of the invention
The technical matters that the present invention mainly solves is to provide a kind of human behavior recognition methods of merging based on multiple features time-space relationship, the method carries out explicit code to the time-space relationship information between the corresponding barycenter of two kinds of features, and the effective information that better can excavate different characteristic carries out human behavior identification.
For solving the problems of the technologies described above, the technical scheme that the present invention adopts is: a kind of human behavior recognition methods of merging based on multiple features time-space relationship, and concrete steps comprise:
Step 1: intensive track characteristic extraction is carried out to video, and with light stream histogram and moving boundaries histogram two kinds of methods, the track characteristic extracted is represented, obtain two kinds of character representations;
Step 2: build the space-time bigraph (bipartite graph) between the corresponding barycenter of two kinds of features with KMEANS algorithm;
Step 3: adopt K road bigraph (bipartite graph) cutting techniques the space-time bigraph (bipartite graph) in step 2 to be divided into the barycenter with strong time-space relationship and the barycenter with weak time-space relationship, the barycenter after segmentation with strong time-space relationship is merged, will there is the barycenter of weak time-space relationship separately;
Step 4: calculate there is strong time-space relationship barycenter between time-space matrix matrix, and adopt the matrix of adjusting the distance of the method for expressing based on conditional probability to compress, obtain the coding of the videl stage after two kinds of Fusion Features;
Step 5: training classifier also identifies.
In a preferred embodiment of the present invention, in described step 2, the two kinds of features obtained in described step 1 are carried out cluster by KMEANS algorithm, thus obtain several barycenter, the time-space relationship between two features is weighed by the L1 distance calculated in each video between the corresponding spacetime coordinates of any two features, utilize the time-space relationship between two kinds of features to calculate time-space relationship between its barycenter, and obtain the space-time bigraph (bipartite graph) between the corresponding barycenter of two kinds of features.
In a preferred embodiment of the present invention, conditional probability described in described step 4 represents that first method carries out discretize to the distance vector between barycenter, then describes any two time-space matrix distributed intelligences of merging between rear barycenter with conditional probability.
The invention has the beneficial effects as follows: a kind of human behavior recognition methods of merging based on multiple features time-space relationship of the present invention, the method is by calculating the time-space matrix in each video between feature, thus the space-time bigraph (bipartite graph) built between the corresponding barycenter of two kinds of features, and adopt K road bigraph (bipartite graph) cutting techniques to split space-time bigraph (bipartite graph), with this, barycenter with strong time-space relationship is merged, the better effective information excavating different characteristic, improves recognition accuracy.
Accompanying drawing explanation
Fig. 1 is a kind of process flow diagram of the human behavior recognition methods based on the fusion of multiple features time-space relationship.
Embodiment
Below preferred embodiment of the present invention is described in detail, can be easier to make advantages and features of the invention be readily appreciated by one skilled in the art, thus more explicit defining is made to protection scope of the present invention.
The embodiment of the present invention comprises: a kind of human behavior recognition methods of merging based on multiple features time-space relationship, and concrete steps comprise:
Step 1: intensive track characteristic extraction is carried out to video, first carries out unique point sampling in an intensive grid.Change of scale can be adapted to make the unique point gathered, to sample in the grid of multiple different spaces yardstick simultaneously, then intensive track characteristic is by estimating that the optical flow field of each frame is followed the tracks of each sampled point, and each sampled point only follows the tracks of L frame in the space scale of its correspondence, finally respectively to feature calculation light stream histogram and moving boundaries histogram as two kinds of different features;
Step 2: establish two kinds of extracting in described step 1 to be characterized as fea1 and fea2, the barycenter that KMEANS cluster obtains is carried out respectively to these two kinds of features and is respectively
with
, construct space-time bigraph (bipartite graph) G (V, E) by the barycenter of these the two kinds of features obtained.
Wherein V=
, E is the adjacency matrix of bigraph (bipartite graph), that is:
Wherein, S is time-space matrix matrix sum between two kinds of features in whole training set, that is:
Wherein,
be the time-space matrix matrix of the corresponding barycenter of two kinds of features in video V, weigh the time-space relationship between two features by the L1 distance calculated in each video between the corresponding spacetime coordinates of any two features, L1 distance computing formula is:
；
Step 3: adopt K road bigraph (bipartite graph) cutting techniques the space-time bigraph (bipartite graph) in step 2 to be divided into the barycenter with strong time-space relationship and the given bigraph (bipartite graph) G (V of barycenter with weak time-space relationship, E), vertex set V is divided into k subset by the K road segmentation of bigraph (bipartite graph)
, object minimizes balancing cost function:
First Laplacian Matrix is constructed
:
And use svd to obtain L's
the left singular vector that individual time singular value is corresponding greatly
with right singular vector
, Qi Zhongshe
,
,
arrive
corresponding L's is second largest to the corresponding left singular vector of the large singular value of l respectively,
arrive
corresponding L's is second largest to the corresponding right singular vector of the large singular value of l respectively;
If
Right
row vector carry out KMEANS cluster, K the barycenter obtained for merge barycenter, if
for the barycenter quantity of feature 1,
for the barycenter quantity of feature 2, then matrix
line number be
, by cluster,
before
barycenter difference character pair 1 belonging to individual row vector
individual barycenter generic,
after
barycenter difference character pair 2 belonging to individual row vector
individual barycenter generic;
Step 4: set merge before video V distance matrix as
, two kinds that carry out merging are characterized as fea1 and fea2, and two kinds of corresponding barycenter set of feature are respectively
with
, and the protoplasm heart to the mapping function merging barycenter is
;
Wherein
,
,
Then the distance matrix of barycenter fusion rear video V is
, wherein
First calculate
fea1barycenter
in video V with
fea2all barycenter with distance
relevant conditional probability is:
Wherein,
represent in video V
fea1barycenter
with
fea2all barycenter apart
number of times,
Then, the barycenter of symmetrical calculating fea2
in video V with
fea1all barycenter with distance
relevant conditional probability is:
Finally, whole video V can be encoded into the matrix of 2m*k:
；
Step 5: last, utilizes the fusion rear video level coding obtained, trains a multi-class support vector machine to be used for the identification of new video.
Compared with prior art, a kind of human behavior recognition methods of merging based on multiple features time-space relationship of the present invention, the method is by calculating the time-space matrix in each video between feature, thus the space-time bigraph (bipartite graph) built between the corresponding barycenter of two kinds of features, and adopt K road bigraph (bipartite graph) cutting techniques to split space-time bigraph (bipartite graph), with this, barycenter with strong time-space relationship is merged, better excavated the effective information of different characteristic, improved recognition accuracy.
The foregoing is only embodiments of the invention; not thereby the scope of the claims of the present invention is limited; every utilize instructions of the present invention to do equivalent structure or the conversion of equivalent flow process, or be directly or indirectly used in other relevant technical fields, be all in like manner included in scope of patent protection of the present invention.
Claims (3)
1., based on the human behavior recognition methods that multiple features time-space relationship merges, it is characterized in that, concrete steps comprise:
Step 1: intensive track characteristic extraction is carried out to video, and with light stream histogram and moving boundaries histogram two kinds of methods, the track characteristic extracted is represented, obtain two kinds of character representations;
Step 2: build the space-time bigraph (bipartite graph) between the corresponding barycenter of two kinds of features with KMEANS algorithm;
Step 3: adopt K road bigraph (bipartite graph) cutting techniques the space-time bigraph (bipartite graph) in step 2 to be divided into the barycenter with strong time-space relationship and the barycenter with weak time-space relationship, the barycenter after segmentation with strong time-space relationship is merged, will there is the barycenter of weak time-space relationship separately;
Step 4: calculate there is strong time-space relationship barycenter between time-space matrix matrix, and adopt the matrix of adjusting the distance of the method for expressing based on conditional probability to compress, obtain the coding of the videl stage after two kinds of Fusion Features;
Step 5: training classifier also identifies.
2. a kind of human behavior recognition methods of merging based on multiple features time-space relationship according to claim 1, it is characterized in that: in described step 2, the two kinds of features obtained in described step 1 are carried out cluster by KMEANS algorithm, thus obtain several barycenter, the time-space relationship between two features is weighed by the L1 distance calculated in each video between the corresponding spacetime coordinates of any two features, utilize the time-space relationship between two kinds of features to calculate time-space relationship between its barycenter, and obtain the space-time bigraph (bipartite graph) between the corresponding barycenter of two kinds of features.
3. a kind of human behavior recognition methods of merging based on multiple features time-space relationship according to claim 1, it is characterized in that: conditional probability described in described step 4 represents that first method carries out discretize to the distance vector between barycenter, then describe any two time-space matrix distributed intelligences of merging between rear barycenter with conditional probability.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510298003.2A CN104881655B (en) | 2015-06-03 | 2015-06-03 | A kind of human behavior recognition methods based on the fusion of multiple features time-space relationship |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510298003.2A CN104881655B (en) | 2015-06-03 | 2015-06-03 | A kind of human behavior recognition methods based on the fusion of multiple features time-space relationship |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104881655A true CN104881655A (en) | 2015-09-02 |
| CN104881655B CN104881655B (en) | 2018-08-28 |
Family
ID=53949142
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510298003.2A Expired - Fee Related CN104881655B (en) | 2015-06-03 | 2015-06-03 | A kind of human behavior recognition methods based on the fusion of multiple features time-space relationship |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104881655B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106056043A (en) * | 2016-05-19 | 2016-10-26 | 中国科学院自动化研究所 | Animal behavior identification method and apparatus based on transfer learning |
| CN106599907A (en) * | 2016-11-29 | 2017-04-26 | 北京航空航天大学 | Multi-feature fusion-based dynamic scene classification method and apparatus |
| CN107680119A (en) * | 2017-09-05 | 2018-02-09 | 燕山大学 | A kind of track algorithm based on space-time context fusion multiple features and scale filter |
| CN108322473A (en) * | 2018-02-12 | 2018-07-24 | 北京京东金融科技控股有限公司 | User behavior analysis method and apparatus |
| CN109508684A (en) * | 2018-11-21 | 2019-03-22 | 中山大学 | A kind of method of Human bodys' response in video |
| CN111080677A (en) * | 2019-12-23 | 2020-04-28 | 天津理工大学 | Protection method for real-time partition operation of workers in pollution remediation site |
| CN111860598A (en) * | 2020-06-18 | 2020-10-30 | 中国地质大学（武汉） | Data analysis method and electronic equipment for identifying sports behaviors and relationships |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103854016A (en) * | 2014-03-27 | 2014-06-11 | 北京大学深圳研究生院 | Human body behavior classification and identification method and system based on directional common occurrence characteristics |
| CN104063721A (en) * | 2014-07-04 | 2014-09-24 | 中国科学院自动化研究所 | Human behavior recognition method based on automatic semantic feature study and screening |
| CN104268577A (en) * | 2014-06-27 | 2015-01-07 | 大连理工大学 | Human body behavior identification method based on inertial sensor |
-
2015
- 2015-06-03 CN CN201510298003.2A patent/CN104881655B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103854016A (en) * | 2014-03-27 | 2014-06-11 | 北京大学深圳研究生院 | Human body behavior classification and identification method and system based on directional common occurrence characteristics |
| CN104268577A (en) * | 2014-06-27 | 2015-01-07 | 大连理工大学 | Human body behavior identification method based on inertial sensor |
| CN104063721A (en) * | 2014-07-04 | 2014-09-24 | 中国科学院自动化研究所 | Human behavior recognition method based on automatic semantic feature study and screening |
Non-Patent Citations (4)
| Title |
|---|
| INDERJIT S.DHILLON等: "Co-clustering documents and words using Bipartite Spectral Graph Partitioning", 《ACM》 * |
| IOSIFIDIS A等: "Discriminant Bag of Words based representation for human action recognition", 《PATTERN RECOGNITION LETTERS》 * |
| JHUO I H等: "Discovering joint audio-visual codewords for video event detection", 《 MACHINE VISION AND APPLICATIONS》 * |
| 王博: "基于时空兴趣点的人体行为识别方法研究", 《中国优秀硕士学位论文全文数据库 信息技术辑》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106056043A (en) * | 2016-05-19 | 2016-10-26 | 中国科学院自动化研究所 | Animal behavior identification method and apparatus based on transfer learning |
| CN106056043B (en) * | 2016-05-19 | 2019-07-30 | 中国科学院自动化研究所 | Animal behavior recognition methods and device based on transfer learning |
| CN106599907A (en) * | 2016-11-29 | 2017-04-26 | 北京航空航天大学 | Multi-feature fusion-based dynamic scene classification method and apparatus |
| CN106599907B (en) * | 2016-11-29 | 2019-11-29 | 北京航空航天大学 | The dynamic scene classification method and device of multiple features fusion |
| CN107680119A (en) * | 2017-09-05 | 2018-02-09 | 燕山大学 | A kind of track algorithm based on space-time context fusion multiple features and scale filter |
| CN108322473A (en) * | 2018-02-12 | 2018-07-24 | 北京京东金融科技控股有限公司 | User behavior analysis method and apparatus |
| CN109508684A (en) * | 2018-11-21 | 2019-03-22 | 中山大学 | A kind of method of Human bodys' response in video |
| CN109508684B (en) * | 2018-11-21 | 2022-12-27 | 中山大学 | Method for recognizing human behavior in video |
| CN111080677A (en) * | 2019-12-23 | 2020-04-28 | 天津理工大学 | Protection method for real-time partition operation of workers in pollution remediation site |
| CN111080677B (en) * | 2019-12-23 | 2023-09-12 | 天津理工大学 | Protection method for real-time partition operation of workers in pollution remediation site |
| CN111860598A (en) * | 2020-06-18 | 2020-10-30 | 中国地质大学（武汉） | Data analysis method and electronic equipment for identifying sports behaviors and relationships |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104881655B (en) | 2018-08-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104881655A (en) | Human behavior recognition method based on multi-feature time-space relationship fusion | |
| CN107563381B (en) | Multi-feature fusion target detection method based on full convolution network | |
| CN105512289B (en) | Image search method based on deep learning and Hash | |
| CN112734775B (en) | Image labeling, image semantic segmentation and model training methods and devices | |
| CN102682302B (en) | Human body posture identification method based on multi-characteristic fusion of key frame | |
| CN105205475A (en) | Dynamic gesture recognition method | |
| CN105069434B (en) | A kind of human action Activity recognition method in video | |
| CN110555387B (en) | Behavior identification method based on space-time volume of local joint point track in skeleton sequence | |
| CN110751674A (en) | Multi-target tracking method and corresponding video analysis system | |
| CN103440645A (en) | Target tracking algorithm based on self-adaptive particle filter and sparse representation | |
| CN112906485A (en) | Visual impairment person auxiliary obstacle perception method based on improved YOLO model | |
| CN105719285A (en) | Pedestrian detection method based on directional chamfering distance characteristics | |
| Wei et al. | P3D-CTN: Pseudo-3D convolutional tube network for spatio-temporal action detection in videos | |
| Chen et al. | CGSANet: A contour-guided and local structure-aware encoder–decoder network for accurate building extraction from very high-resolution remote sensing imagery | |
| Liu et al. | Towards interpretable and robust hand detection via pixel-wise prediction | |
| CN113298817A (en) | High-accuracy semantic segmentation method for remote sensing image | |
| CN108573241A (en) | A kind of video behavior recognition methods based on fusion feature | |
| Lu et al. | MFNet: Multi-feature fusion network for real-time semantic segmentation in road scenes | |
| CN115457082A (en) | Pedestrian multi-target tracking algorithm based on multi-feature fusion enhancement | |
| Li et al. | Robust detection of headland boundary in paddy fields from continuous RGB-D images using hybrid deep neural networks | |
| CN105956604A (en) | Action identification method based on two layers of space-time neighborhood characteristics | |
| Ouyang et al. | Aerial target detection based on the improved YOLOv3 algorithm | |
| CN113780140A (en) | Gesture image segmentation and recognition method and device based on deep learning | |
| Zhou et al. | Boundary-guided lightweight semantic segmentation with multi-scale semantic context | |
| CN113255569A (en) | 3D attitude estimation method based on image hole convolutional encoder decoder |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180828 Termination date: 20190603 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |