CN108596054A - A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features - Google Patents
A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features Download PDFInfo
- Publication number
- CN108596054A CN108596054A CN201810314275.0A CN201810314275A CN108596054A CN 108596054 A CN108596054 A CN 108596054A CN 201810314275 A CN201810314275 A CN 201810314275A CN 108596054 A CN108596054 A CN 108596054A
- Authority
- CN
- China
- Prior art keywords
- characteristic pattern
- crowd
- density
- counting method
- people counting
- 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
- 238000000034 method Methods 0.000 title claims abstract description 27
- 230000004927 fusion Effects 0.000 title claims abstract description 10
- 238000012549 training Methods 0.000 claims description 6
- 241000287196 Asthenes Species 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims 1
- 230000002547 anomalous effect Effects 0.000 abstract description 3
- 238000012544 monitoring process Methods 0.000 abstract description 3
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- 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
Abstract
The invention discloses a kind of people counting methods based on multiple dimensioned full convolutional network Fusion Features comprising following steps:A pictures are inputted, three branching networks are respectively enterd, for obtaining the characteristic pattern of different scale;The characteristic pattern of each branching networks of acquisition is merged, the characteristic pattern finally estimated;The characteristic pattern of output is mapped as density map;The estimation that summation operation realizes current crowd's quantity is carried out to density map.A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features provided by the invention, can overcome block, the distortion of scene perspective and different Crowds Distribute etc. seriously affect crowd's counting, with stronger practicability, robustness is good, crowd's counting or crowd density estimation can be accurately carried out, there will be important value to monitoring crowd's anomalous event.
Description
Technical field
The present invention is to be related to a kind of people counting method based on multiple dimensioned full convolutional network Fusion Features, is belonged at image
Manage technical field.
Background technology
Population size estimation in intensive occasion has many potential actual application values, including monitoring (for example, detection is abnormal big
Crowded crowd, or control the quantity of people in an area), (record enters or leaves some region for safety management
Number), urban planning (for example, analyzing the flow of the people in some region) etc..
The method counted in the prior art about crowd mainly has:
1) pedestrian detection method:In the scene of crowd's sparse distribution, counted by detecting each pedestrian in video.
This method is relatively direct, but can be affected in crowded block.
2) trajectory clustering method:For monitor video, usually using KLT (Kanade-Lucas-Tomasi) trackers and gather
The method of class removes estimated number with trajectory clustering.But this method can be limited by serious block between people.
3) the feature Return Law:Characteristics of image and the regression model of number are obtained first, are then estimated by characteristics of image
Number in image.This method considers the global feature of crowd, can carry out large-scale crowd's counting, but can neglect
The slightly spatial information of pedestrian, causes counting precision inadequate.
4) the crowd density figure Return Law:By the target density figure regression count of Pixel-level, target is study minutia
Mapping between crowd's counting.The method can estimate the target numbers of image any position, it is proposed that interactive object count
System.But cross-scenario counting is not particularly suited for both for special scenes at present.
In short, due to blocking, crowd's irregular distribution the problems such as, cause dense population counting to be still faced with very big challenge,
Therefore, research and develop it is a kind of it is highly practical, robustness is good, can accurately carry out crowd's counting or crowd density estimation method will to prison
Control crowd's anomalous event is of great significance.
Invention content
In view of the above-mentioned problems existing in the prior art and demand, the object of the present invention is to provide a kind of highly practical, robusts
The property people counting method based on multiple dimensioned full convolutional network Fusion Features that is good, can accurately carrying out crowd's counting.
To achieve the above object, the technical solution adopted by the present invention is as follows:
A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features, includes the following steps:
S1:A pictures are inputted, three branching networks are respectively enterd, for obtaining the characteristic pattern of different scale;
S2:The characteristic pattern of each branching networks of acquisition is merged, the characteristic pattern finally estimated;
S3:The characteristic pattern of output is mapped as density map;
S4:The estimation that summation operation realizes current crowd's quantity is carried out to density map.
Furtherly, the characteristic pattern of the different scale described in step S1 is different size of due to being used in branching networks
Convolution kernel and the network number of plies, which differ, to be caused.
Furtherly, the concrete operations of step S1 are:
A) picture of the head position label with someone is first converted into crowd density figure, if there is a head position exists
Pixel xi, it is denoted as δ (x-xi), then there is the image of the head position label of N number of people to be represented by functional expression (1):
B) by functional expression (1) and Gaussian kernel GσConvolution is carried out, density estimation functional expression (2) is obtained:
F (x)=H (x) * Gσ(x) (2)
C) everyone propagation parameter σ is automatically determined according to the average distance data with its neighbour people, if by given figure
The distance of every head to its k nearest neighbours as in are expressed asThen average distance is represented by letter
Numerical expression (3):
D) by δ (x-xi) and propagation parameter σiWith average distance daCarry out convolution to proportional Gaussian kernel, then it is accurately close
Degree F is represented by functional expression (4):
Parameter beta therein is to check label using the geometric self-adaptation for being suitable for the local geometric around each data point is interior
H carries out convolution and obtains.
Preferably, β=0.3.
Furtherly, in the picture input convolutional neural networks described in step S1, when defining the loss function of network,
Joint training is carried out using two loss functions, specific two loss functions definition is as shown in functional expression (5) and (6):
In above-mentioned formula:θ is one group of parameter that can learn, and N is the quantity of training image, XiRepresenting input images, FiIt is image
XiGround real density figure, F (Xi;θ) it is network-evaluated density map, ziRepresentative image XiIn true crowd's quantity, s generations
Table estimates the area of space of density map, and the L in functional expression (5) is the damage estimated between density map and ground real density figure
It loses, the L in functional expression (6) is the loss between the true number of estimated number and ground.
Furtherly, the concrete operations that the characteristic pattern of each branching networks by acquisition described in step S2 merges
It is:So that the tensor of input data is obtained corresponding input tensor after branching networks, keep row dimension constant, to row dimension into
Row connection.
Furtherly, the specific method that the characteristic pattern of output is mapped as to density map described in step S3 is special in output
The convolution operation that primary convolution kernel size is 1 × 1 is carried out on sign figure.
Compared with prior art, the present invention has the advantages that:
A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features provided by the invention, can overcome screening
Gear, the distortion of scene perspective and different Crowds Distributes etc. seriously affect crowd's counting, have stronger practicability, robustness
It is good, crowd's counting or crowd density estimation can be accurately carried out, there will be important value to monitoring crowd's anomalous event.
Description of the drawings
Fig. 1 is estimation density map of the embodiment of the present invention to crowd's picture.
Specific implementation mode
Technical scheme of the present invention is described in further detail below in conjunction with drawings and examples.
Embodiment
A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features provided in this embodiment, including it is following
Step:
S1:A pictures are inputted, three branching networks are respectively enterd, for obtaining the characteristic pattern of different scale;Specifically
Operation is:
A) picture of the head position label with someone is first converted into crowd density figure, if there is a head position exists
Pixel xi, it is denoted as δ (x-xi), then there is the image of the head position label of N number of people to be represented by functional expression (1):
B) by functional expression (1) and Gaussian kernel GσConvolution is carried out, density estimation functional expression (2) is obtained:
F (x)=H (x) * Gσ(x) (2)
C) everyone propagation parameter σ is automatically determined according to the average distance data with its neighbour people, if by given figure
The distance of every head to its k nearest neighbours as in are expressed asThen average distance is represented by letter
Numerical expression (3):
D) by δ (x-xi) and propagation parameter σiWith average distance daCarry out convolution to proportional Gaussian kernel, then it is accurately close
Degree F is represented by functional expression (4):
Parameter beta therein is to check label using the geometric self-adaptation for being suitable for the local geometric around each data point is interior
H carries out convolution and obtains.The experimental results showed that:β=0.3 can obtain best result.
In the picture input convolutional neural networks, when defining the loss function of network, using two loss functions
Joint training is carried out, specific two loss functions definition is as shown in functional expression (5) and (6):
In above-mentioned formula:θ is one group of parameter that can learn, and N is the quantity of training image, XiRepresenting input images, FiIt is image
XiGround real density figure, F (Xi;θ) it is network-evaluated density map, ziRepresentative image XiIn true crowd's quantity, s generations
Table estimates the area of space of density map, and the L in functional expression (5) is the damage estimated between density map and ground real density figure
It loses, the L in functional expression (6) is the loss between the true number of estimated number and ground.
S2:The characteristic pattern of each branching networks of acquisition is merged, the characteristic pattern finally estimated;Concrete operations
It is:So that the tensor of input data is obtained corresponding input tensor after branching networks, keep row dimension constant, to row dimension into
Row connection.
S3:The characteristic pattern of output is mapped as density map, specific method is to carry out primary convolution on output characteristic pattern
The convolution operation that core size is 1 × 1.
S4:The estimation that summation operation realizes current crowd's quantity is carried out to density map.
It is last it is necessarily pointed out that:The foregoing is merely the preferable specific implementation mode of the present invention, but the present invention
Protection domain be not limited thereto, any one skilled in the art in the technical scope disclosed by the present invention,
The change or replacement that can be readily occurred in, should be covered by the protection scope of the present invention.
Claims (7)
1. a kind of people counting method based on multiple dimensioned full convolutional network Fusion Features, which is characterized in that include the following steps:
S1:A pictures are inputted, three branching networks are respectively enterd, for obtaining the characteristic pattern of different scale;
S2:The characteristic pattern of each branching networks of acquisition is merged, the characteristic pattern finally estimated;
S3:The characteristic pattern of output is mapped as density map;
S4:The estimation that summation operation realizes current crowd's quantity is carried out to density map.
2. people counting method according to claim 1, it is characterised in that:The feature of different scale described in step S1
Figure is caused due to being differed using different size of convolution kernel and the network number of plies in branching networks.
3. people counting method according to claim 1, which is characterized in that the concrete operations of step S1 are:
A) picture of the head position label with someone is first converted into crowd density figure, if there is a head position is in pixel
Point xi, it is denoted as δ (x-xi), then there is the image of the head position label of N number of people to be represented by functional expression (1):
B) by functional expression (1) and Gaussian kernel GσConvolution is carried out, density estimation functional expression (2) is obtained:
F (x)=H (x) * Gσ(x) (2)
C) everyone propagation parameter σ is automatically determined according to the average distance data with its neighbour people, if by given image
The distance of every head to its k neighbours recently be expressed asThen average distance is represented by functional expression
(3):
D) by δ (x-xi) and propagation parameter σiWith average distance daConvolution is carried out to proportional Gaussian kernel, then accurate density F
It is represented by functional expression (4):
Parameter beta therein be using be suitable in the geometric self-adaptation of the local geometric around each data point verification label H into
Row convolution obtains.
4. people counting method according to claim 3, it is characterised in that:β=0.3.
5. people counting method according to claim 1, it is characterised in that:Picture input convolution god described in step S1
Through in network, when defining the loss function of network, joint training, specific two losses letter are carried out using two loss functions
Number definition is as shown in functional expression (5) and (6):
In above-mentioned formula:θ is one group of parameter that can learn, and N is the quantity of training image, XiRepresenting input images, FiIt is image Xi's
Ground real density figure, F (Xi;θ) it is network-evaluated density map, ZiRepresentative image XiIn true crowd's quantity, S representatives estimate
The area of space of density map is counted, and the L in functional expression (5) is the loss estimated between density map and ground real density figure, letter
L in numerical expression (6) is the loss between the true number of estimated number and ground.
6. people counting method according to claim 1, which is characterized in that each point by acquisition described in step S2
The concrete operations that the characteristic pattern of branch network is merged are:The tensor of input data is set to be obtained after branching networks corresponding defeated
Enter tensor, keeps row dimension constant, row dimension is attached.
7. people counting method according to claim 1, it is characterised in that:The characteristic pattern by output described in step S3
The specific method for being mapped as density map is that the convolution operation that primary convolution kernel size is 1 × 1 is carried out on output characteristic pattern.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810314275.0A CN108596054A (en) | 2018-04-10 | 2018-04-10 | A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810314275.0A CN108596054A (en) | 2018-04-10 | 2018-04-10 | A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108596054A true CN108596054A (en) | 2018-09-28 |
Family
ID=63621480
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810314275.0A Pending CN108596054A (en) | 2018-04-10 | 2018-04-10 | A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108596054A (en) |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109271960A (en) * | 2018-10-08 | 2019-01-25 | 燕山大学 | A kind of demographic method based on convolutional neural networks |
CN109447008A (en) * | 2018-11-02 | 2019-03-08 | 中山大学 | Population analysis method based on attention mechanism and deformable convolutional neural networks |
CN109492615A (en) * | 2018-11-29 | 2019-03-19 | 中山大学 | Crowd density estimation method based on CNN low layer semantic feature density map |
CN109614941A (en) * | 2018-12-14 | 2019-04-12 | 中山大学 | A kind of embedded crowd density estimation method based on convolutional neural networks model |
CN109815919A (en) * | 2019-01-28 | 2019-05-28 | 上海七牛信息技术有限公司 | A kind of people counting method, network, system and electronic equipment |
CN109858424A (en) * | 2019-01-25 | 2019-06-07 | 佳都新太科技股份有限公司 | Crowd density statistical method, device, electronic equipment and storage medium |
CN110263849A (en) * | 2019-06-19 | 2019-09-20 | 合肥工业大学 | A kind of crowd density estimation method based on multiple dimensioned attention mechanism |
CN110598558A (en) * | 2019-08-14 | 2019-12-20 | 浙江省北大信息技术高等研究院 | Crowd density estimation method, device, electronic equipment and medium |
CN111027389A (en) * | 2019-11-12 | 2020-04-17 | 通号通信信息集团有限公司 | Training data generation method based on deformable Gaussian kernel in crowd counting system |
CN111062274A (en) * | 2019-12-02 | 2020-04-24 | 汇纳科技股份有限公司 | Context-aware embedded crowd counting method, system, medium, and electronic device |
CN111753671A (en) * | 2020-06-02 | 2020-10-09 | 华东师范大学 | Crowd counting method for real scene |
CN112101164A (en) * | 2020-09-06 | 2020-12-18 | 西北工业大学 | Lightweight crowd counting method based on full convolution network |
CN112115900A (en) * | 2020-09-24 | 2020-12-22 | 腾讯科技(深圳)有限公司 | Image processing method, device, equipment and storage medium |
CN112215129A (en) * | 2020-10-10 | 2021-01-12 | 江南大学 | Crowd counting method and system based on sequencing loss and double-branch network |
CN112597985A (en) * | 2021-03-04 | 2021-04-02 | 成都西交智汇大数据科技有限公司 | Crowd counting method based on multi-scale feature fusion |
CN112632601A (en) * | 2020-12-16 | 2021-04-09 | 苏州玖合智能科技有限公司 | Crowd counting method for subway carriage scene |
CN112767316A (en) * | 2020-12-31 | 2021-05-07 | 山东师范大学 | Crowd counting method and system based on multi-scale interactive network |
CN113221971A (en) * | 2021-04-25 | 2021-08-06 | 山东师范大学 | Multi-scale crowd counting method and system based on front and back feature fusion |
CN113538400A (en) * | 2021-07-29 | 2021-10-22 | 燕山大学 | Cross-modal crowd counting method and system |
CN112115900B (en) * | 2020-09-24 | 2024-04-30 | 腾讯科技(深圳)有限公司 | Image processing method, device, equipment and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105528589A (en) * | 2015-12-31 | 2016-04-27 | 上海科技大学 | Single image crowd counting algorithm based on multi-column convolutional neural network |
CN107341480A (en) * | 2017-07-12 | 2017-11-10 | 中国电子科技集团公司第二十八研究所 | A kind of crowd massing detection method of modified PCCNN neural network models |
US20180005071A1 (en) * | 2013-06-25 | 2018-01-04 | University Of Central Florida Research Foundation, Inc. | Multi-Source, Multi-Scale Counting in Dense Crowd Images |
CN107742099A (en) * | 2017-09-30 | 2018-02-27 | 四川云图睿视科技有限公司 | A kind of crowd density estimation based on full convolutional network, the method for demographics |
-
2018
- 2018-04-10 CN CN201810314275.0A patent/CN108596054A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20180005071A1 (en) * | 2013-06-25 | 2018-01-04 | University Of Central Florida Research Foundation, Inc. | Multi-Source, Multi-Scale Counting in Dense Crowd Images |
CN105528589A (en) * | 2015-12-31 | 2016-04-27 | 上海科技大学 | Single image crowd counting algorithm based on multi-column convolutional neural network |
CN107341480A (en) * | 2017-07-12 | 2017-11-10 | 中国电子科技集团公司第二十八研究所 | A kind of crowd massing detection method of modified PCCNN neural network models |
CN107742099A (en) * | 2017-09-30 | 2018-02-27 | 四川云图睿视科技有限公司 | A kind of crowd density estimation based on full convolutional network, the method for demographics |
Non-Patent Citations (2)
Title |
---|
YINGYING ZHANG ET.AL: "Single-Image Crowd Counting via Multi-Column Convolutional Neural Network", 《2016 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
彭山珍 等: "基于多尺度全卷积网络特征融合的人群计数", 《武汉大学学报(理学版)》 * |
Cited By (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109271960A (en) * | 2018-10-08 | 2019-01-25 | 燕山大学 | A kind of demographic method based on convolutional neural networks |
CN109447008A (en) * | 2018-11-02 | 2019-03-08 | 中山大学 | Population analysis method based on attention mechanism and deformable convolutional neural networks |
CN109492615A (en) * | 2018-11-29 | 2019-03-19 | 中山大学 | Crowd density estimation method based on CNN low layer semantic feature density map |
CN109614941A (en) * | 2018-12-14 | 2019-04-12 | 中山大学 | A kind of embedded crowd density estimation method based on convolutional neural networks model |
CN109614941B (en) * | 2018-12-14 | 2023-02-03 | 中山大学 | Embedded crowd density estimation method based on convolutional neural network model |
CN109858424A (en) * | 2019-01-25 | 2019-06-07 | 佳都新太科技股份有限公司 | Crowd density statistical method, device, electronic equipment and storage medium |
CN109815919A (en) * | 2019-01-28 | 2019-05-28 | 上海七牛信息技术有限公司 | A kind of people counting method, network, system and electronic equipment |
CN109815919B (en) * | 2019-01-28 | 2021-04-20 | 上海七牛信息技术有限公司 | Crowd counting method, network, system and electronic equipment |
CN110263849A (en) * | 2019-06-19 | 2019-09-20 | 合肥工业大学 | A kind of crowd density estimation method based on multiple dimensioned attention mechanism |
CN110598558A (en) * | 2019-08-14 | 2019-12-20 | 浙江省北大信息技术高等研究院 | Crowd density estimation method, device, electronic equipment and medium |
CN111027389A (en) * | 2019-11-12 | 2020-04-17 | 通号通信信息集团有限公司 | Training data generation method based on deformable Gaussian kernel in crowd counting system |
CN111027389B (en) * | 2019-11-12 | 2023-06-30 | 通号通信信息集团有限公司 | Training data generation method based on deformable Gaussian kernel in crowd counting system |
CN111062274A (en) * | 2019-12-02 | 2020-04-24 | 汇纳科技股份有限公司 | Context-aware embedded crowd counting method, system, medium, and electronic device |
CN111062274B (en) * | 2019-12-02 | 2023-11-28 | 汇纳科技股份有限公司 | Context-aware embedded crowd counting method, system, medium and electronic equipment |
CN111753671A (en) * | 2020-06-02 | 2020-10-09 | 华东师范大学 | Crowd counting method for real scene |
CN112101164A (en) * | 2020-09-06 | 2020-12-18 | 西北工业大学 | Lightweight crowd counting method based on full convolution network |
CN112115900A (en) * | 2020-09-24 | 2020-12-22 | 腾讯科技(深圳)有限公司 | Image processing method, device, equipment and storage medium |
CN112115900B (en) * | 2020-09-24 | 2024-04-30 | 腾讯科技(深圳)有限公司 | Image processing method, device, equipment and storage medium |
CN112215129A (en) * | 2020-10-10 | 2021-01-12 | 江南大学 | Crowd counting method and system based on sequencing loss and double-branch network |
CN112632601A (en) * | 2020-12-16 | 2021-04-09 | 苏州玖合智能科技有限公司 | Crowd counting method for subway carriage scene |
CN112632601B (en) * | 2020-12-16 | 2024-03-12 | 苏州玖合智能科技有限公司 | Crowd counting method for subway carriage scene |
CN112767316A (en) * | 2020-12-31 | 2021-05-07 | 山东师范大学 | Crowd counting method and system based on multi-scale interactive network |
CN112597985A (en) * | 2021-03-04 | 2021-04-02 | 成都西交智汇大数据科技有限公司 | Crowd counting method based on multi-scale feature fusion |
CN113221971A (en) * | 2021-04-25 | 2021-08-06 | 山东师范大学 | Multi-scale crowd counting method and system based on front and back feature fusion |
CN113221971B (en) * | 2021-04-25 | 2022-12-27 | 山东师范大学 | Multi-scale crowd counting method and system based on front and back feature fusion |
CN113538400A (en) * | 2021-07-29 | 2021-10-22 | 燕山大学 | Cross-modal crowd counting method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108596054A (en) | A kind of people counting method based on multiple dimensioned full convolutional network Fusion Features | |
CN110135269B (en) | Fire image detection method based on mixed color model and neural network | |
CN108830145B (en) | People counting method based on deep neural network and storage medium | |
Van Etten et al. | The multi-temporal urban development spacenet dataset | |
Han et al. | Fast saliency-aware multi-modality image fusion | |
CN108665487B (en) | Transformer substation operation object and target positioning method based on infrared and visible light fusion | |
CN107679503A (en) | A kind of crowd's counting algorithm based on deep learning | |
JP2019505866A (en) | Passerby head identification method and system | |
CN109241871A (en) | A kind of public domain stream of people's tracking based on video data | |
CN111209892A (en) | Crowd density and quantity estimation method based on convolutional neural network | |
CN104978567B (en) | Vehicle checking method based on scene classification | |
JP2009533778A (en) | Video segmentation using statistical pixel modeling | |
CN103577875A (en) | CAD (computer-aided design) people counting method based on FAST (features from accelerated segment test) | |
Hinz et al. | Car detection in aerial thermal images by local and global evidence accumulation | |
CN108154110A (en) | A kind of intensive people flow amount statistical method based on the detection of the deep learning number of people | |
CN110210379A (en) | A kind of lens boundary detection method of combination critical movements feature and color characteristic | |
CN103489012B (en) | Crowd density detecting method and system based on support vector machine | |
Patil et al. | Motion saliency based generative adversarial network for underwater moving object segmentation | |
CN111783589A (en) | Complex scene crowd counting method based on scene classification and multi-scale feature fusion | |
CN109948683A (en) | Difficulty division methods, device and its relevant device of point cloud data | |
Tomar et al. | Crowd analysis in video surveillance: A review | |
Zhang et al. | New mixed adaptive detection algorithm for moving target with big data | |
Colombari et al. | Patch-based background initialization in heavily cluttered video | |
CN110930384A (en) | Crowd counting method, device, equipment and medium based on density information | |
CN108280842B (en) | Foreground segmentation method for overcoming illumination mutation |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180928 |