CN110084197A - Bus passenger flow volume statistical method and system based on deep learning - Google Patents
Bus passenger flow volume statistical method and system based on deep learning Download PDFInfo
- Publication number
- CN110084197A CN110084197A CN201910347705.3A CN201910347705A CN110084197A CN 110084197 A CN110084197 A CN 110084197A CN 201910347705 A CN201910347705 A CN 201910347705A CN 110084197 A CN110084197 A CN 110084197A
- Authority
- CN
- China
- Prior art keywords
- target
- people
- deep learning
- passenger flow
- sample data
- 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
- 238000013135 deep learning Methods 0.000 title claims abstract description 28
- 238000007619 statistical method Methods 0.000 title claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 36
- 238000001514 detection method Methods 0.000 claims abstract description 26
- 238000012360 testing method Methods 0.000 claims abstract description 14
- 238000010200 validation analysis Methods 0.000 claims abstract description 12
- 238000013136 deep learning model Methods 0.000 claims abstract description 6
- 238000007689 inspection Methods 0.000 claims abstract description 5
- 238000012545 processing Methods 0.000 claims description 24
- 238000000034 method Methods 0.000 claims description 20
- 238000010586 diagram Methods 0.000 claims description 12
- 230000001629 suppression Effects 0.000 claims description 7
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 230000002708 enhancing effect Effects 0.000 claims description 6
- 238000005286 illumination Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000005259 measurement Methods 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims 1
- 238000006243 chemical reaction Methods 0.000 abstract 1
- 238000005516 engineering process Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
Abstract
The invention discloses a kind of bus passenger flow volume statistical methods based on deep learning, including acquire passenger flow sample data offline, row format of going forward side by side conversion；Deep learning model is built to be trained；The model obtained to training is tested using test sample, saves the model parameter learnt；Live video stream is detected using the model learnt, and stores the number of people information detected；The number of people information that continuous multiple frames detect is tracked, determines multiple moving targets；Determining moving target is compared with preset count threshold, the counting when moving target reaches count threshold.In number of people detection, secondary verifying is carried out to the target frame detected, it is input in trained sorter network in advance, class validation is carried out to the target frame, it whether is the number of people, it greatly improves the recall rate of the number of people and reduces missing inspection, the false detection rate of the number of people, the problems such as very good solution congestion, knapsack, band cap.
Description
Technical field
The present invention relates to a kind of counting passenger flow of buses systems, more particularly to a kind of bus based on deep learning
Passenger flow volume statistical method and system.
Background technique
Guest flow statistics technology common at present mainly has following 3 kinds:
(1) artificial statistical method
It is manually recorded to the boarding and alighting progress of each website that artificial statistics mainly makees personnel by trip runner, and in terminal
Stand and carry out manual collect statistics, this method advantage is accuracy rate height, but person works' amount it is bigger and can not real-time interconnection exist
Line counts the real-time boarding and alighting of each website.
(2) it is based on IC card method
IC card stores a large amount of personal information of passenger, and the ID of each passenger that swipes the card of this method record carries out passenger flow statistics, and advantage is
Without arranging extra work personnel to count with vehicle, but accuracy rate is influenced bigger, and IC by IC card utilization rate
Card is not networked, and cannot achieve the online real-time statistics of vehicle-mounted passenger flow.
(3) volume of the flow of passengers automatic counting method based on image vision
This method is combined by GPS positioning technology, network transmission technology and image vision processing technique realizes that public transit vehicle is each
The real-time programming count of a website passenger getting on/off time and number.Wherein image vision processing technique is mainly to the number of people of getting on or off the bus
Carry out detecting and tracking, realize the automatic counting of the volume of the flow of passengers, but the accuracy rate of traditional machine learning image processing method by
Light influence is bigger, and can not solve the problems such as crowded, with cap, knapsack.The present invention is therefore.
Summary of the invention
In order to solve the above-mentioned technical problem, the invention proposes a kind of counting passenger flow of buses sides based on deep learning
Method and system carry out secondary verifying to the target frame detected, are input to trained sorter network in advance in number of people detection
In, class validation is carried out to the target frame, whether is the number of people, the recall rate of the number of people is greatly improved and reduces the number of people
Missing inspection, false detection rate, the problems such as very good solution congestion, knapsack, band cap.
The technical scheme adopted by the invention is that:
A kind of bus passenger flow volume statistical method based on deep learning, comprising the following steps:
S01: offline acquisition passenger flow sample data, and sample data is divided into trained and test sample two parts, and be converted to use
In the format of caffe deep learning frame；
S02: building passenger flow number of people deep learning model, is trained using sample data by caffe deep learning frame；
S03: the model obtained to training is tested using test sample, if the recall rate of model measurement is lower than given threshold,
Then continue training pattern, conversely, obtaining final model parameter；
S04: detecting live video stream using the obtained model of study, obtain target block diagram picture, and stores and detect
Number of people information；
S05: the number of people information that continuous multiple frames detect is tracked, determines multiple moving targets；
S06: determining human body target is compared with preset count threshold, when moving target reaches count threshold
It counts.
In preferred technical solution, the step S04 further includes carrying out secondary classification verifying to obtained target block diagram picture,
Include:
S41: constructing the convolutional neural networks of lightweight, uses single phase training method end to end；
S42: convolution pond Processing with Neural Network is carried out to pretreated target block diagram picture, obtains feature vector chart；
S43: it in the feature figure layer of number of people target detection, is waited respectively in the target that each point constructs multiple and different scale sizes
Select frame；
S44: classification is carried out to target candidate frame and bounding box recurrence is handled, obtains target number of people position；
S45: to obtained target number of people position, non-maxima suppression processing is carried out, target position is obtained；
S46: secondary verifying is carried out to the target frame detected, is input in trained sorter network in advance, to the target frame
Class validation is carried out, the number of people is judged whether it is.
In preferred technical solution, further includes abundant sample data in the step S01, be included in various illumination, cap
Original training sample data are acquired in the case of form, complex background；The raw sample data of acquisition is carried out in training pattern
Data enhancing processing overturns image, is rotated, being translated, being cut, scaling adjustment, addition Gaussian noise processing, and adjustment
The brightness of image, saturation degree, contrast.
In preferred technical solution, if further including having detected door signal triggering in the step S06, statistics is final
The number of getting on or off the bus upload to net background.
The counting passenger flow of buses system based on deep learning that the invention also discloses a kind of, comprising:
Offline sample data acquisition processing module, it is offline to acquire passenger flow sample data, and sample data is divided into training and test
Sample two parts, and it is converted to the format for caffe deep learning frame；
Model construction module builds passenger flow number of people deep learning model, using sample data by caffe deep learning frame into
Row training；
Training module, the model obtained to training are tested using test sample, if the recall rate of model measurement is lower than setting
Threshold value then continues training pattern, conversely, obtaining final model parameter；
On-line checking module detects live video stream using the model learnt, obtains target block diagram picture, and store inspection
The number of people information measured；
Tracking module tracks the number of people information that continuous multiple frames detect, determines multiple moving targets；
Determining moving target is compared by counting module with preset count threshold, when moving target reaches counting
It is counted when threshold value.
In preferred technical solution, the on-line checking module further includes class validation module, the class validation module
Secondary verifying is carried out to the target block diagram picture detected, comprising:
S41: constructing the convolutional neural networks of lightweight, uses single phase training method end to end；
S42: convolution pond Processing with Neural Network is carried out to pretreated target block diagram picture, obtains feature vector chart；
S43: it in the feature figure layer of number of people target detection, is waited respectively in the target that each point constructs multiple and different scale sizes
Select frame；
S44: classification is carried out to target candidate frame and bounding box recurrence is handled, obtains target number of people position；
S45: to obtained target number of people position, non-maxima suppression processing is carried out, target position is obtained；
S46: secondary verifying is carried out to the target frame detected, is input in trained sorter network in advance, to the target frame
Class validation is carried out, the number of people is judged whether it is.
In preferred technical solution, the offline sample data acquisition processing module further includes abundant sample data module,
It is included in the case of various illumination, cap form, complex background and acquires original training sample data；In training pattern to acquisition
Raw sample data carry out data enhancing processing, image is overturn, is rotated, is translated, is cut, scale adjustment, adds height
This noise processed, and brightness, the saturation degree, contrast of adjustment image.
In preferred technical solution, if further including having detected door signal triggering in the counting module, count most
The whole number of getting on or off the bus uploads to net background.
Compared with prior art, the beneficial effects of the present invention are:
1, when the number of people detects, secondary verifying is carried out to the target frame detected, is input in trained sorter network in advance,
Class validation is carried out to the target frame, whether is the number of people, passes through extensive, multi-pose, various illumination, complex background data
Study automatically extract correlated characteristic, target number of people internal information abundant can be portrayed, greatly improve the detection of the number of people
The problems such as rate and missing inspection, false detection rate for reducing the number of people, very good solution congestion, knapsack, band cap.
, in the case where detecting the target number of people, by target tracking algorism track target complex in target moved
Target is counted in conjunction with threshold line, to realize the accurate metering of the volume of the flow of passengers.
, by the Background statistic of continuous multiple frames image, can be realized the judgement blocked to camera.Method of the invention at
This is lower, more efficient.
Detailed description of the invention
The invention will be further described with reference to the accompanying drawings and embodiments:
Fig. 1 is that the present invention is based on the flow charts of the bus passenger flow volume statistical method of deep learning；
Specific embodiment.
In order to make the objectives, technical solutions and advantages of the present invention clearer, With reference to embodiment and join
According to attached drawing, the present invention is described in more detail.It should be understood that these descriptions are merely illustrative, and it is not intended to limit this hair
Bright range.In addition, in the following description, descriptions of well-known structures and technologies are omitted, to avoid this is unnecessarily obscured
The concept of invention.
Embodiment
As shown in Figure 1, including the following steps: the present invention is based on the bus passenger flow volume statistical method of deep learning
Off-line training passenger flow number of people detection-phase includes the steps that being A1, A2, A3, and mainly training obtains the ideal people of effect
Head detection model parameter, the on-line checking for on-line stage；The online passenger flow statistics stage in real time, include the steps that for A4, A5,
A6, A7 carry out on-line checking, passenger flow statistics.
: offline acquisition passenger flow sample data is divided into trained and test sample two parts, and be converted to caffe deep learning
The LMDB format that frame needs.
: build passenger flow number of people deep learning model, be trained based on sample data by caffe deep learning frame and
Test, study obtain final model parameter file.
: the model obtained to A2 stage-training is tested using test sample, is wanted if the recall rate of model measurement is lower than
It asks, then continues training pattern, if reaching target, the model parameter learnt is saved, so that this model is used for on-line stage
It carries out actually detected.
: this stage mainly judges camera with the presence or absence of circumstance of occlusion, when having detected opening signal triggering,
Algorithm can statistically analyze camera video stream several seconds, if continuous multiple frames image detection is judged as camera shooting less than multidate information
Head blocks, and exports and block exception, and multidate information then carries out the detection in A5 stage if it exists.
: this stage mainly detects live video stream using the model that off-line phase learns, and stores detection
The number of people information arrived.
: this stage mainly tracks the number of people target information that continuous multiple frames detect, determines multiple moving targets.
: this stage is mainly to analyze and determine to the stage A6 moving target determined and preset threshold line, when
It is then counted automatically when moving target passes through threshold line；If having detected door signal triggering, the final number of getting on or off the bus of statistics
Upload to net background.
, it is difficult for detection in A6, A7 and be easy the knapsack of error, problem of attaching the names of pre-determined candidates, be based on depth learning technology, design one kind
The convolutional neural networks of lightweight are distinguished using single phase training method end to end by the characteristic pattern to different convolutional layers
It is detected, result is finally integrated by detection number of people position using non-maxima suppression method, compared to traditional method,
Both algorithm Detection accuracy had been improved, the real-time of algorithm is also improved.While in order to make detection model adapt to band cap, knapsack etc.
Various forms scene increases the robustness of model, carries out greatly abundant enhancing to the data sample of training pattern parameter, mainly
By following approach, original training data sample is acquired in various illumination, cap form, complex background；In training mould
Various data enhancings is done to collected primary data sample when type to handle, and such as image is overturn, rotate, translate, is cut out
Various adjustment, addition Gaussian noise processing, and brightness, the saturation degree, contrast of adjustment image such as cut, scale.These methods
The sample for enriching training pattern makes network architecture can be very good study and arrives the numbers of people features such as the number of people, cap, knapsack,
Improve the precision of the detection target number of people.
In terms of model structure, the Feature Mapping of the different scale of image is extracted by CNN network, in the convolution of different depth
Layer prediction number of people target, and a part overlapping or incorrect target person are curbed by non-maximum suppression (NMS) method
Head has lower confidence level for being easy other targets that erroneous detection is the number of people, such as knapsack, circular object when generally detecting
Threshold value can screen out these pseudo- targets by the way that reasonable confidence threshold value is arranged, finally obtain target number of people position Rect (x,
y, width, height)。
Number of people detection method end to end based on deep learning, specifically includes the following steps:
1) pretreatment such as compressed, normalized to the whole picture for being input to network model；
2) convolution pond Processing with Neural Network is carried out to pretreated picture, obtains the feature vector charts such as gradient, edge；
3) in the feature figure layer of number of people target detection, respectively in the target candidate that each puts 6 different scale sizes of construction
Frame；
4) softmax classification is carried out to target candidate frame and bounding box recurrence is handled, obtain true number of people target position；
5) for obtained target number of people position, NMS(non-maxima suppression is carried out) processing, generate final number of people testing result
Rect (x, y, width, height), x, y, central point x, y-coordinate, width object height, height object width.
In number of people detection, secondary verifying is carried out to the target frame detected, is input to trained sorter network in advance
In, class validation is carried out to the target frame, whether is the number of people.
The method of the present invention has detection effect well for crowded, band cap situation, does not have erroneous detection for the case where knapsack,
More accurate number of people information is provided for subsequent statistical counting.
It should be understood that above-mentioned specific embodiment of the invention is used only for exemplary illustration or explains of the invention
Principle, but not to limit the present invention.Therefore, that is done without departing from the spirit and scope of the present invention is any
Modification, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.In addition, appended claims purport of the present invention
Covering the whole variations fallen into attached claim scope and boundary or this range and the equivalent form on boundary and is repairing
Change example.
Claims (8)
1. a kind of bus passenger flow volume statistical method based on deep learning, which comprises the following steps:
S01: offline acquisition passenger flow sample data, and sample data is divided into trained and test sample two parts, and be converted to use
In the format of caffe deep learning frame；
S02: building passenger flow number of people deep learning model, is trained using sample data by caffe deep learning frame；
S03: the model obtained to training is tested using test sample, if the recall rate of model measurement is lower than given threshold,
Then continue training pattern, conversely, obtaining final model parameter；
S04: detecting live video stream using the obtained model of study, obtain target block diagram picture, and stores and detect
Number of people information；
S05: the number of people information that continuous multiple frames detect is tracked, determines multiple moving targets；
S06: determining human body target is compared with preset count threshold, when moving target reaches count threshold
It counts.
2. the bus passenger flow volume statistical method according to claim 1 based on deep learning, which is characterized in that the step
Rapid S04 further includes carrying out secondary classification verifying to obtained target block diagram picture, comprising:
S41: constructing the convolutional neural networks of lightweight, uses single phase training method end to end；
S42: convolution pond Processing with Neural Network is carried out to pretreated target block diagram picture, obtains feature vector chart；
S43: it in the feature figure layer of number of people target detection, is waited respectively in the target that each point constructs multiple and different scale sizes
Select frame；
S44: classification is carried out to target candidate frame and bounding box recurrence is handled, obtains target number of people position；
S45: to obtained target number of people position, non-maxima suppression processing is carried out, target position is obtained；
S46: secondary verifying is carried out to the target frame detected, is input in trained sorter network in advance, to the target frame
Class validation is carried out, the number of people is judged whether it is.
3. the bus passenger flow volume statistical method according to claim 1 based on deep learning, which is characterized in that the step
Further include abundant sample data in rapid S01, is included in the case of various illumination, cap form, complex background and acquires original training
Sample data；Data enhancing processing is carried out to the raw sample data of acquisition in training pattern, image is overturn, is revolved
Turn, translation, cut, scaling adjustment, addition Gaussian noise processing, and brightness, the saturation degree, contrast of adjustment image.
4. the bus passenger flow volume statistical method according to claim 1 based on deep learning, which is characterized in that the step
If further including having detected door signal triggering in rapid S06, the final number of getting on or off the bus of statistics uploads to net background.
5. a kind of counting passenger flow of buses system based on deep learning characterized by comprising
Offline sample data acquisition processing module, it is offline to acquire passenger flow sample data, and sample data is divided into training and test
Sample two parts, and it is converted to the format for caffe deep learning frame；
Model construction module builds passenger flow number of people deep learning model, using sample data by caffe deep learning frame into
Row training；
Training module, the model obtained to training are tested using test sample, if the recall rate of model measurement is lower than setting
Threshold value then continues training pattern, conversely, obtaining final model parameter；
On-line checking module detects live video stream using the model learnt, obtains target block diagram picture, and store inspection
The number of people information measured；
Tracking module tracks the number of people information that continuous multiple frames detect, determines multiple moving targets；
Determining moving target is compared by counting module with preset count threshold, when moving target reaches counting
It is counted when threshold value.
6. the counting passenger flow of buses system according to claim 5 based on deep learning, which is characterized in that it is described
Line detection module further includes class validation module, and the class validation module carries out secondary test to the target block diagram picture detected
Card, comprising:
S41: constructing the convolutional neural networks of lightweight, uses single phase training method end to end；
S42: convolution pond Processing with Neural Network is carried out to pretreated target block diagram picture, obtains feature vector chart；
S43: it in the feature figure layer of number of people target detection, is waited respectively in the target that each point constructs multiple and different scale sizes
Select frame；
S44: classification is carried out to target candidate frame and bounding box recurrence is handled, obtains target number of people position；
S45: to obtained target number of people position, non-maxima suppression processing is carried out, target position is obtained；
S46: secondary verifying is carried out to the target frame detected, is input in trained sorter network in advance, to the target frame
Class validation is carried out, the number of people is judged whether it is.
7. the counting passenger flow of buses system according to claim 5 based on deep learning, which is characterized in that it is described from
Line sample data acquisition processing module further includes abundant sample data module, is included in various illumination, cap form, complex background
In the case of acquire original training sample data；Data enhancing processing is carried out to the raw sample data of acquisition in training pattern,
Image is overturn, is rotated, is translated, is cut, scales adjustment, addition Gaussian noise processing, and adjusts the brightness of image, satisfy
With degree, contrast.
8. the counting passenger flow of buses system according to claim 5 based on deep learning, which is characterized in that the meter
If further including having detected door signal triggering in digital-to-analogue block, the final number of getting on or off the bus of statistics uploads to net background.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910347705.3A CN110084197A (en) | 2019-04-28 | 2019-04-28 | Bus passenger flow volume statistical method and system based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910347705.3A CN110084197A (en) | 2019-04-28 | 2019-04-28 | Bus passenger flow volume statistical method and system based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110084197A true CN110084197A (en) | 2019-08-02 |
Family
ID=67417213
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910347705.3A Pending CN110084197A (en) | 2019-04-28 | 2019-04-28 | Bus passenger flow volume statistical method and system based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110084197A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110516602A (en) * | 2019-08-28 | 2019-11-29 | 杭州律橙电子科技有限公司 | A kind of public traffice passenger flow statistical method based on monocular camera and depth learning technology |
| CN110555419A (en) * | 2019-09-09 | 2019-12-10 | 江苏慧眼数据科技股份有限公司 | passenger flow counting method based on binocular stereo vision |
| CN110992305A (en) * | 2019-10-31 | 2020-04-10 | 中山大学 | Package counting method and system based on deep learning and multi-target tracking technology |
| CN111257865A (en) * | 2020-02-07 | 2020-06-09 | 电子科技大学 | Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model |
| CN111382720A (en) * | 2020-03-18 | 2020-07-07 | 浩云科技股份有限公司 | Real-time personnel number statistical method and device based on top-view monocular camera |
| CN111507762A (en) * | 2020-04-15 | 2020-08-07 | 中国科学院自动化研究所 | Urban taxi demand prediction method based on multi-task co-prediction neural network |
| CN111723673A (en) * | 2020-05-25 | 2020-09-29 | 西安交通大学 | Intelligent high-speed rail people counting method based on computer vision |
| CN112215809A (en) * | 2020-09-25 | 2021-01-12 | 北京林业大学 | Prawn processing shelling rate detection method and device, electronic equipment and storage medium |
| CN112541374A (en) * | 2019-09-20 | 2021-03-23 | 南京行者易智能交通科技有限公司 | Passenger attribute acquisition method and device based on deep learning and model training method |
| CN112633210A (en) * | 2020-12-14 | 2021-04-09 | 南京理工大学 | Rail transit passenger flow density estimation system and method based on target detection |
| CN112633057A (en) * | 2020-11-04 | 2021-04-09 | 北方工业大学 | Intelligent monitoring method for abnormal behaviors in bus |
| CN112686090A (en) * | 2020-11-04 | 2021-04-20 | 北方工业大学 | Intelligent monitoring system for abnormal behaviors in bus |
| CN113177445A (en) * | 2021-04-16 | 2021-07-27 | 新华智云科技有限公司 | Video mirror moving identification method and system |
| CN114140735A (en) * | 2021-07-12 | 2022-03-04 | 苏州万店掌网络科技有限公司 | Deep learning-based goods path accumulation detection method and system and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101847265A (en) * | 2010-04-20 | 2010-09-29 | 上海理工大学 | Method for extracting moving objects and partitioning multiple objects used in bus passenger flow statistical system |
| CN103646253A (en) * | 2013-12-16 | 2014-03-19 | 重庆大学 | Bus passenger flow statistics method based on multi-motion passenger behavior analysis |
| CN105426867A (en) * | 2015-12-11 | 2016-03-23 | 小米科技有限责任公司 | Face identification verification method and apparatus |
| CN107239762A (en) * | 2017-06-06 | 2017-10-10 | 电子科技大学 | Patronage statistical method in a kind of bus of view-based access control model |
| US20180144204A1 (en) * | 2016-03-17 | 2018-05-24 | Nec Corporation | Passenger counting device, system, method and program, and vehicle movement amount calculation device, method and program |
| CN108416250A (en) * | 2017-02-10 | 2018-08-17 | 浙江宇视科技有限公司 | Demographic method and device |
| CN109325404A (en) * | 2018-08-07 | 2019-02-12 | 长安大学 | A kind of demographic method under public transport scene |
-
2019
- 2019-04-28 CN CN201910347705.3A patent/CN110084197A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101847265A (en) * | 2010-04-20 | 2010-09-29 | 上海理工大学 | Method for extracting moving objects and partitioning multiple objects used in bus passenger flow statistical system |
| CN103646253A (en) * | 2013-12-16 | 2014-03-19 | 重庆大学 | Bus passenger flow statistics method based on multi-motion passenger behavior analysis |
| CN105426867A (en) * | 2015-12-11 | 2016-03-23 | 小米科技有限责任公司 | Face identification verification method and apparatus |
| US20180144204A1 (en) * | 2016-03-17 | 2018-05-24 | Nec Corporation | Passenger counting device, system, method and program, and vehicle movement amount calculation device, method and program |
| CN108416250A (en) * | 2017-02-10 | 2018-08-17 | 浙江宇视科技有限公司 | Demographic method and device |
| CN107239762A (en) * | 2017-06-06 | 2017-10-10 | 电子科技大学 | Patronage statistical method in a kind of bus of view-based access control model |
| CN109325404A (en) * | 2018-08-07 | 2019-02-12 | 长安大学 | A kind of demographic method under public transport scene |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110516602A (en) * | 2019-08-28 | 2019-11-29 | 杭州律橙电子科技有限公司 | A kind of public traffice passenger flow statistical method based on monocular camera and depth learning technology |
| CN110555419A (en) * | 2019-09-09 | 2019-12-10 | 江苏慧眼数据科技股份有限公司 | passenger flow counting method based on binocular stereo vision |
| CN110555419B (en) * | 2019-09-09 | 2023-05-26 | 江苏慧眼数据科技股份有限公司 | Passenger flow counting method based on binocular stereoscopic vision |
| CN112541374A (en) * | 2019-09-20 | 2021-03-23 | 南京行者易智能交通科技有限公司 | Passenger attribute acquisition method and device based on deep learning and model training method |
| CN110992305A (en) * | 2019-10-31 | 2020-04-10 | 中山大学 | Package counting method and system based on deep learning and multi-target tracking technology |
| CN111257865A (en) * | 2020-02-07 | 2020-06-09 | 电子科技大学 | Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model |
| CN111382720A (en) * | 2020-03-18 | 2020-07-07 | 浩云科技股份有限公司 | Real-time personnel number statistical method and device based on top-view monocular camera |
| CN111507762A (en) * | 2020-04-15 | 2020-08-07 | 中国科学院自动化研究所 | Urban taxi demand prediction method based on multi-task co-prediction neural network |
| CN111507762B (en) * | 2020-04-15 | 2023-10-31 | 中国科学院自动化研究所 | Urban taxi demand prediction method based on multitasking co-prediction neural network |
| CN111723673A (en) * | 2020-05-25 | 2020-09-29 | 西安交通大学 | Intelligent high-speed rail people counting method based on computer vision |
| CN112215809A (en) * | 2020-09-25 | 2021-01-12 | 北京林业大学 | Prawn processing shelling rate detection method and device, electronic equipment and storage medium |
| CN112633057A (en) * | 2020-11-04 | 2021-04-09 | 北方工业大学 | Intelligent monitoring method for abnormal behaviors in bus |
| CN112686090A (en) * | 2020-11-04 | 2021-04-20 | 北方工业大学 | Intelligent monitoring system for abnormal behaviors in bus |
| CN112633057B (en) * | 2020-11-04 | 2024-01-30 | 北方工业大学 | Intelligent monitoring method for abnormal behavior in bus |
| CN112686090B (en) * | 2020-11-04 | 2024-02-06 | 北方工业大学 | Intelligent monitoring system for abnormal behavior in bus |
| CN112633210A (en) * | 2020-12-14 | 2021-04-09 | 南京理工大学 | Rail transit passenger flow density estimation system and method based on target detection |
| CN113177445A (en) * | 2021-04-16 | 2021-07-27 | 新华智云科技有限公司 | Video mirror moving identification method and system |
| CN114140735A (en) * | 2021-07-12 | 2022-03-04 | 苏州万店掌网络科技有限公司 | Deep learning-based goods path accumulation detection method and system and storage medium |
| CN114140735B (en) * | 2021-07-12 | 2023-04-18 | 苏州万店掌网络科技有限公司 | Deep learning-based goods path accumulation detection method and system and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110084197A (en) | Bus passenger flow volume statistical method and system based on deep learning | |
| CN107563347B (en) | Passenger flow counting method and device based on TOF camera | |
| CN108351967A (en) | A kind of plurality of human faces detection method, device, server, system and storage medium | |
| CN105373135B (en) | A kind of method and system of aircraft docking guidance and plane type recognition based on machine vision | |
| CN103699905B (en) | Method and device for positioning license plate | |
| CN110487562A (en) | One kind being used for unpiloted road-holding ability detection system and method | |
| US20110176000A1 (en) | System and Method for Counting People | |
| CN103077526B (en) | There is train method for detecting abnormality and the system of depth detection function | |
| CN109145708B (en) | Pedestrian flow statistical method based on RGB and D information fusion | |
| JPH11175730A (en) | Human body detection and trace system | |
| CN106611160A (en) | CNN (Convolutional Neural Network) based image hair identification method and device | |
| CN110255318B (en) | Method for detecting idle articles in elevator car based on image semantic segmentation | |
| CN110111565A (en) | A kind of people's vehicle flowrate System and method for flowed down based on real-time video | |
| CN107273852A (en) | Escalator floor plates object and passenger behavior detection algorithm based on machine vision | |
| CN110427815B (en) | Video processing method and device for realizing interception of effective contents of entrance guard | |
| CN109598947A (en) | A kind of vehicle identification method and system | |
| CN108345878B (en) | Public transport passenger flow monitoring method and system based on video | |
| CN107798688A (en) | Motion estimate method, method for early warning and automobile anti-rear end collision prior-warning device | |
| CN111079589A (en) | Automatic height detection method based on depth camera shooting and height threshold value pixel calibration | |
| CN110728269B (en) | High-speed rail contact net support pole number plate identification method based on C2 detection data | |
| CN108932464A (en) | Passenger flow volume statistical method and device | |
| CN112597965A (en) | Driving behavior recognition method and device and computer readable storage medium | |
| CN109344829B (en) | Train number identification method and device for high-speed railway train | |
| CN110516600A (en) | A kind of bus passenger flow detection method based on Face datection | |
| CN108154199B (en) | High-precision rapid single-class target detection method based on deep learning |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190802 |