CN103413114A - Near-drowning behavior detection method based on support vector machine - Google Patents
Near-drowning behavior detection method based on support vector machine Download PDFInfo
- Publication number
- CN103413114A CN103413114A CN2013101890951A CN201310189095A CN103413114A CN 103413114 A CN103413114 A CN 103413114A CN 2013101890951 A CN2013101890951 A CN 2013101890951A CN 201310189095 A CN201310189095 A CN 201310189095A CN 103413114 A CN103413114 A CN 103413114A
- Authority
- CN
- China
- Prior art keywords
- swimmer
- behavior
- support vector
- vector machine
- state
- 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
Images
Abstract
The invention discloses a near-drowning behavior detection method based on a support vector machine. According to the invention, the support vector machine is used as a classifier for the learning of a machine; the support vector machine classifier is trained through an obtained video sequence sample of near-drowning behavior and normal swimming behavior by pre-simulation; then a video image sequence of a pool is acquired in real time through a camera arranged above the water; the monitored video image sequence is inputted into the trained support vector machine classifier to determine a behavioral state of a swimmer. Therefore, a near-drowner can be automatically detected through the camera in an actual public swimming place and lives can be timely saved at the maximum. The near-drowning behavior detection method based on the support vector machine has the advantages of accurate and reliable detection, good robustness, high noise immunity and good adaption to transformation of light. Besides, through monitoring by the camera arranged above the water, the near-drowning behavior detection method based on the support vector machine lowers costs for system implementation and has great value in engineering application.
Description
Technical field
The present invention relates to computer vision technique and field of video monitoring, relate in particular to a kind of drowned behavior detection method based on support vector machine.
Background technology
Swimming pool is the capital construction facility in each city, is the main place of people's amusement and recreation.Yet according to a data, show, drowning is the first cause of China teenager unexpected death.Many drowning incidents all occur in public swimming place, even professional lifesaving personnel supervision is arranged, but find in time the drowning person owing to reaching, and have caused the generation of death incident.Therefore, research and development are a set of can find to detect the system of drowning incident automatically to find that in time the drowning person saves life and has great realistic meaning.
Before more than 20 years, carried out the research of the automatic context of detection of drowned behavior abroad.In prior art, some technical scheme proposes, by ultrasonic array is installed, to detect the movement rate of swimmer's strokes, as the signal that may find drowned behavior; Also have some technical schemes to propose to utilize ultrasonic radar, detect the swimming pool bottom and whether occur not having the object of motion to judge whether to have occurred drowning incident; The Poseidon Technology of France has developed in the world first set based on computer assisted drowned behavior detection system in addition.This system, by above water-bed and the water surface, camera being installed, detects the behavior of analyzing crowd in swimming pool, if note abnormalities the involved party, namely the drowning person, just give the alarm by display and alarm bell.This system is installed application for many years in the public swimming place of a plurality of countries and regions of America and Europe, and helps to find the multiple row drowning person, has saved many people's life.But also there are many drawbacks in this system, such as, this system can only detect the drowning person of drowned later stage of discovery, drowning person's life now is precarious, simultaneously, the passing through of this system is arranged on water-bed camera and realizes monitoring, and such scheme not only maintenance cost is high, and the accuracy rate of monitoring is easy to be subjected to the interference of other objects.The application number that the applicant applied on Dec 28th, 2011 is called the patent of invention of " based on the early stage drowned behavior act detection method of video " for " 201110448257.X " name, this patent discloses a kind of drowned behavior detection method that adopts Hidden Markov Model (HMM), but the method is trained and modeler model process complexity in earlier stage, real-time bad in practical implementation, the accuracy detected simultaneously is not high, and false drop rate is high.
Therefore, for the deficiency that current drowned behavior detection method exists, the present invention proposes a kind of efficient stable simply based on the drowned behavior detection method of support vector machine.
Summary of the invention
The objective of the invention is to overcome the deficiency of the drowned behavior detection technique of existing employing camera, a kind of drowned behavior detection method based on support vector machine is provided.
The objective of the invention is to be achieved through the following technical solutions: a kind of drowned behavior detection method based on support vector machine, the method comprises following content:
(1) sequence of video images of the camera Real-time Collection swimming pool by being arranged on water surface top;
(2) adopt the foreground moving target in the sequence of video images obtained in the background subtraction method extraction step (1) based on code book, i.e. swimmer;
(3) to the swimmer who obtains in step (2), adopt the method for ellipse fitting to set up manikin, calculating is based on characteristics of human body's parameter sets Q of this model, and characteristics of human body's parameter sets Q comprises angle Qpos, minimum external matrix area ratio Qact, area change amount Qsv, the underwater portion ratio Qsub of axis and horizontal axis;
(4) the trained support vector machine classifier of the parameter of the characteristics of human body about swimmer input step (3) obtained, judge the state that this swimmer is in;
(5) swimmer's who step (4) is obtained current state is recorded in state recording sequence R;
(6) adopt the swimmer who detects in Kalman filtering algorithm tracking step (2);
(7) repeating step (2)-(6) several times, the result of this swimmer's of keeping records several times state-detection is in state recording sequence R;
(8) by the state recording sequence R and the state behavior relation table that in step (7), obtain, judge that this swimmer is in normal swimming behavior or potential drowned behavior, if the judgement behavior be potential drowned behavior, by current potential drowned behavior record in behavior records series T; If the behavior of judgement is normal swimming behavior, abandon behavior record;
(9) in behavior record sequence T, there is potential drowned behavior in some continuous times, judge that the swimmer is in drowned behavior, gives the alarm.
The invention has the beneficial effects as follows, the present invention is enough automatically to be detected and finds the drowning person by camera in the public swimming of reality place, can save in time to greatest extent life, have advantages of detection accurately and reliably, robustness is good, noiseproof feature is high, to the swimming pool good environmental adaptability; And the present invention is by being arranged on the camera implementing monitoring of water surface top, the System Implementation cost is low, has great engineering using value.
The accompanying drawing explanation
The background model that Fig. 1 is based on the code book algorithm builds schematic diagram;
Fig. 2 is ellipse fitting manikin schematic diagram;
Fig. 3 is swimmer's state definition schematic diagram.
Embodiment
Below in conjunction with accompanying drawing, describe the present invention in detail, it is more obvious that purpose of the present invention and effect will become.The drowned behavior detection method that the present invention is based on support vector machine comprises the steps:
Step 1: the sequence of video images of the camera Real-time Collection swimming pool by being arranged on water surface top.
The camera that the inventive method gathers the image employing is the ordinary video monitoring camera, obtains the image size and is the 352*288 sequence of video images.
Step 2: adopt the foreground moving target in the sequence of video images obtained in the background subtraction method extraction step 1 based on code book, i.e. the swimmer.The concrete execution step of background subtraction method based on code book is as follows:
2.1) obtain the RGB image of sequence of video images;
2.2) for step 2.1) and in each pixel in the RGB image that obtains set up code book
, wherein
For forming the code word of code book,
,
lFor the number of code word in code book, code word
By 26 n dimensional vector ns of rgb color vector, formed, wherein, 6 n dimensional vector ns comprise brightness value maximum in brightness value minimum in code word, code word, the frequency that code word occurs, the passive distance of swimming of maximum of code word, time and the last time occurred of code word that code word occurs for the first time;
2.3) empty background that some frames are not comprised to the swimmer is as the reference frame of training background model, calculates respectively in these images the code book of the correspondence of each pixel in every two field picture, the set of these code books built to background model by as shown in Figure 1 mode, wherein
Be illustrated in background model,
(i, j)The dictionary that the pixel of position is corresponding, this dictionary is by the code book of the reference frame correspondence position from for background modeling
Form, wherein
(i, j)Mean code book
The corresponding position of pixel in image,
nMean the
nThe frame reference frame;
2.4) calculate the code book corresponding to each pixel of video image of input;
2.5) take pixel and be unit, by step 2.4) in code book and the background model of the input picture that obtains mate, if this pixel finds the code book of coupling in the relevant position of background model, this pixel is background pixel, otherwise this pixel is the foreground target image;
2.6) by step 2.5) image after processing carries out binary conversion treatment, image after binary conversion treatment is carried out to the Image erosion operation and remove the noise in image, by the image closed procedure, repair the foveola and the border interruption that in foreground target, exist and isolate final swimmer.
Step 3: adopt the method for ellipse fitting to set up manikin to the swimmer who obtains in step 2, calculating is based on characteristics of human body's parameter sets Q of this model, and characteristics of human body's parameter sets Q comprises angle Qpos, minimum external matrix area ratio Qact, area change amount Qsv and the underwater portion ratio Qsub of axis and horizontal axis.
In the patent " based on the early stage drowned behavior act detection method of video " of inventor's application in 2011, mention the employing ellipse and carried out Human Modeling, but in the method, do not explicitly pointed out the rule of oval modeling, therefore can not carry out Accurate Model to human body.The present invention proposes employing method as shown in Figure 2 to Human Modeling, with 5 ellipses, mean respectively head, health, left hand, the right hand, left leg and the right leg of human body.Each ellipse comprises axis coordinate, minor axis coordinate and 3 parameters of focal coordinates.
For the video image obtained in step 1 is set up coordinate system, each characteristics of human body's parameter is defined as follows in this coordinate system:
3.1) the variable angle amount (parameter Qpos) of axis and horizontal axis characterizes the axis of foreground target and the variation of the angle of horizontal axis, the irregular row moved due to the drowning person, so compare drowning person's Q with normal swimmer
posBe worth larger.
3.2) minimum external matrix area ratio (parameter Qact) is defined as the minimum external matrix of foreground target at a period of time inner area ratio.Normal swimmer is because the trick campaign is regular, and Qact can be a relatively stable smaller value, and the action of drowning person's trick is generally disorderly and unsystematic, thus Qact be one relatively unstable, change large value.
3.3) area change amount (parameter Qsv) characterizes the variation of foreground target at a period of time inner area, namely
Wherein
,
,
Be illustrated respectively in maximal value, minimum value and the mean value of the area of a period of time interval foreground target.The irregular movement range of drowning person's trick is large, and therefore relatively normal its area change of swimmer is larger.
3.4) the underwater situation of underwater portion ratio (parameter Qsub) sign foreground target submergence.When prospect target health was got over many parts and is immersed in the water, the color saturation of foreground target can be higher, and therefore, parameter Qsub is defined as in a period of time interval, the difference of foreground target full color saturation and minimum color saturation.The part that the drowning person surfaces in the struggle stage in early days is fewer, is generally head or trick, and therefore, with respect to normal swimmer, the drowning person has larger Qsub value.
Step 4: the trained support vector machine classifier of the parameter of the characteristics of human body about swimmer input by step 3 obtains judges the state that this swimmer is in.
The present invention adopts support vector machine classifier, the characteristics of human body's parameter during by training in advance learning swimmer different conditions, thus realize the Classification and Identification to swimmer's Different activity states.Wherein the training study process of support vector machine classifier is as follows:
4.1) by engaging professional performer, imitate drowned behavior and normal swimming behavior, by the camera that is arranged on water surface top, take the process that record imitates, obtain the sequence of video images sample;
4.2) by step 4.1) and in the action Definition of Division of swimmer in every frame in the sequence of video images that obtains be 4 kinds of states, as shown in Figure 3, respectively the swimmer state s1 parallel with the water surface, the state s4 that the state s3 that the state s2 of swimmer and horizontal vertical, swimmer are tilted to the right in the water surface and swimmer are tilted to the left in level;
4.3) according to step 2 extraction step 4.1) swimmer in the video sequence sample that obtains, then be designated as respectively Q1, Q2, Q3 and Q4 according to characteristics of human body's parameter that step 3 is calculated 4 kinds of states of swimmer;
4.4) by step 4.3) and in the Q1 that obtains, Q2, Q3 and Q4 characteristics of human body parameter input support vector machine classifier, complete training study.
Step 5: the swimmer's that step (4) is obtained current state is recorded in state recording sequence R.
The state recording sequence R related in this step is the array of 10 yuan:
, wherein
,
,
For step 4.2) middle 4 kinds of swimmer's states that define.
Step 6: adopt the swimmer who detects in Kalman filtering algorithm tracking step 2.
Step 7: repeating step (2)-(6) several times, the result of this swimmer's of keeping records several times state-detection is in state recording sequence R.
Step 8: by state recording sequence R and the state behavior relation table obtained in step (7), judge that this swimmer is in normal swimming behavior or potential drowned behavior, if the behavior of judgement is potential drowned behavior, and by current potential drowned behavior record in behavior records series T; If the behavior of judgement is normal swimming behavior, abandon behavior record.
State behavior relation table is as shown in table 1, and the number percent in figure is each state shared number percent in several times state record table R.By the operating state of swimmer in continuous time being detected to judge the behavior situation of swimmer in current slot, the rule of its judgement is as follows:
8.1) if in several times state recording R, swimmer's plumbness surpasses 80%, judges that the swimmer is potential drowned behavior;
8.2) in several times state recording R, except the rule 8.1) other situations be judged as normal swimming behavior;
Step 9: when there being potential drowned behavior in 3 continuous times in behavior record sequence T, judge that the swimmer is in drowned behavior, gives the alarm.
Table 1: state behavior relation table
The present invention adopts the sorter of support vector machine as machine learning, by simulation, obtain in advance the video sequence sample training support vector machine classifier of drowned behavior and normal swimming behavior, then the sequence of video images of the camera Real-time Collection swimming pool by being arranged on water surface top, the support vector machine classifier that the input of the sequence of video images that monitors is trained, judgement swimmer's behavior state, thereby can by camera, automatically detect and find the drowning person in the public swimming of reality place, can save in time to greatest extent life, has detection accurately and reliably, robustness is good, noiseproof feature is high, advantage to the swimming pool good environmental adaptability, and the present invention is by being arranged on the camera implementing monitoring of water surface top, the System Implementation cost is low, has great engineering using value.
Claims (7)
1. the drowned behavior detection method based on support vector machine, is characterized in that, the method comprises following content:
(1) sequence of video images of the camera Real-time Collection swimming pool by being arranged on water surface top;
(2) adopt the foreground moving target in the sequence of video images obtained in the background subtraction method extraction step (1) based on code book, i.e. swimmer;
(3) to the swimmer who obtains in step (2), adopt the method for ellipse fitting to set up manikin, calculating is based on characteristics of human body's parameter sets Q of this model, and these characteristics of human body's parameters comprise angle Qpos, minimum external matrix area ratio Qact, area change amount Qsv, the underwater portion ratio Qsub of axis and horizontal axis;
(4) the trained support vector machine classifier of the parameter of the characteristics of human body about swimmer input step (3) obtained, judge the state that this swimmer is in;
(5) swimmer's who step (4) is obtained current state is recorded in state recording sequence R;
(6) adopt the swimmer who detects in Kalman filtering algorithm tracking step (2);
(7) repeating step (2)-(6) several times, the result of this swimmer's of keeping records several times state-detection is in state recording sequence R;
(8) by the state recording sequence R and the state behavior relation table that in step (7), obtain, judge that this swimmer is in normal swimming behavior or potential drowned behavior, if the behavior of judgement is potential drowned behavior, and by current potential drowned behavior record in behavior records series T; If the behavior of judgement is normal swimming behavior, abandon behavior record;
(9) in behavior record sequence T, there is potential drowned behavior in some continuous times, judge that the swimmer is in drowned behavior, gives the alarm.
2. the drowned behavior detection method based on support vector machine according to claim 1, is characterized in that, in described step 2, as follows based on the concrete execution step of background subtraction method of code book:
(2.1) obtain the RGB image of sequence of video images;
(2.2) for step 2.1) in each pixel in the RGB image that obtains set up code book
, wherein,
For forming the code word of code book,
,
lFor the number of code word in code book, code word
By 26 n dimensional vector ns of rgb color vector, formed, wherein, 6 n dimensional vector ns comprise brightness value maximum in brightness value minimum in code word, code word, the frequency that code word occurs, the passive distance of swimming of maximum of code word, time and the last time occurred of code word that code word occurs for the first time;
(2.3) some frames are not comprised to the reference frame of swimmer's empty background as the training background model, calculate respectively in these images the code book of the correspondence of each pixel in every two field picture, the set of these code books is built to background model by as shown in Figure 1 mode, wherein
Be illustrated in background model
(i, j)The dictionary that the pixel of position is corresponding, this dictionary is by the code book of the reference frame correspondence position from for background modeling
Form, wherein
(i, j)Mean code book
The corresponding position of pixel in image,
nMean the
nThe frame reference frame;
(2.4) calculate the code book corresponding to each pixel of video image of input;
(2.5) take pixel is unit, and code book and the background model of the input picture that obtains in step 2.4 are mated, if this pixel finds the code book of coupling in the relevant position of background model, this pixel is background pixel, otherwise this pixel is the foreground target image;
(2.6) image after step 2.5 processing is carried out to binary conversion treatment, image after binary conversion treatment is carried out to the Image erosion operation and remove the noise in image, by the image closed procedure, repair the foveola and the border interruption that in foreground target, exist and isolate final swimmer.
3. the drowned behavior detection method based on support vector machine according to claim 1, it is characterized in that, in described step 3, the method for described employing ellipse fitting human body is: the head, health, left hand, the right hand, left leg and the right leg that with 5 ellipses, mean respectively human body; Each ellipse comprises axis coordinate, minor axis coordinate and 3 parameters of focal coordinates.
4. the drowned behavior detection method based on support vector machine according to claim 1, is characterized in that, in described step 3, each characteristics of human body's parameter is defined as follows:
(3.1) variation of the angle of the axis of the variable angle amount Qpos of axis and horizontal axis sign foreground target and horizontal axis, due to the irregular row of drowning person's action, so with normal swimmer, compare drowning person's Q
posBe worth larger;
(3.2) minimum external matrix area ratio Qact is defined as the minimum external matrix of foreground target at a period of time inner area ratio; Normal swimmer is because the trick campaign is regular, and Qact can be a relatively stable smaller value, and the action of drowning person's trick is generally disorderly and unsystematic, thus Qact be one relatively unstable, change large value;
(3.3) area change amount Qsv characterizes the variation of foreground target at a period of time inner area, namely
Wherein
,
,
Be illustrated respectively in maximal value, minimum value and the mean value of the area of a period of time interval foreground target; The irregular movement range of drowning person's trick is large, and therefore relatively normal its area change of swimmer is larger;
(3.4) underwater portion ratio Qsub characterizes the underwater situation of foreground target submergence, when prospect target health is got over many parts and is immersed in the water, the color saturation of foreground target can be higher, therefore, parameter Qsub is defined as in a period of time interval, the difference of foreground target full color saturation and minimum color saturation; The part that the drowning person surfaces in the struggle stage in early days is fewer, is generally head or trick, and therefore, with respect to normal swimmer, the drowning person has larger Qsub value.
5. the drowned behavior detection method based on support vector machine according to claim 1, is characterized in that, in described step 4, the training study process of support vector machine classifier is as follows:
(4.1) by engaging professional performer to imitate drowned behavior and normal swimming behavior, take by the camera that is arranged on the water surface top process that record imitates, obtain the sequence of video images sample;
(4.2) the action Definition of Division in every frame is 4 kinds of states by the swimmer in the sequence of video images obtained in step 4.1, as shown in Figure 3, respectively the swimmer state s1 parallel with the water surface, the state s4 that the state s3 that the state s2 of swimmer and horizontal vertical, swimmer are tilted to the right in the water surface and swimmer are tilted to the left in level;
(4.3) swimmer in the video sequence sample obtained according to step 2 extraction step 4.1, then be designated as respectively Q1, Q2, Q3 and Q4 according to characteristics of human body's parameter that step 3 is calculated 4 kinds of states of swimmer;
(4.4) by step 4.3) in the Q1 that obtains, Q2, Q3 and Q4 characteristics of human body parameter input support vector machine classifier, complete training study.
7. the drowned behavior detection method based on support vector machine according to claim 1, is characterized in that, in described step 8, state behavior relation table is as shown in the table:
Number percent in table is each state shared number percent in several times state record table R; By the operating state of swimmer in continuous time being detected to judge the behavior situation of swimmer in current slot, the rule of its judgement is as follows:
(8.1) if in several times state recording R, swimmer's plumbness surpasses 80%, judges that the swimmer is potential drowned behavior;
(8.2) in several times state recording R, other situations except rule 8.1 are judged as normal swimming behavior.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2013101890951A CN103413114A (en) | 2013-05-17 | 2013-05-17 | Near-drowning behavior detection method based on support vector machine |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2013101890951A CN103413114A (en) | 2013-05-17 | 2013-05-17 | Near-drowning behavior detection method based on support vector machine |
Publications (1)
Publication Number | Publication Date |
---|---|
CN103413114A true CN103413114A (en) | 2013-11-27 |
Family
ID=49606123
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN2013101890951A Pending CN103413114A (en) | 2013-05-17 | 2013-05-17 | Near-drowning behavior detection method based on support vector machine |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103413114A (en) |
Cited By (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104902229A (en) * | 2015-05-19 | 2015-09-09 | 吴晗 | Video monitoring method, system and camera shooting monitoring system |
CN105187764A (en) * | 2015-06-26 | 2015-12-23 | 哈尔滨工程大学 | Underwater measurement lifesaving device based on visual analysis |
CN106022230A (en) * | 2016-05-11 | 2016-10-12 | 太原理工大学 | Video-based detection method for drowning event in swimming pool |
CN106570854A (en) * | 2015-10-09 | 2017-04-19 | 富士通株式会社 | Water state detection device and method and image processing device |
WO2017130187A1 (en) * | 2016-01-26 | 2017-08-03 | Coral Detection Systems Ltd. | Methods and systems for drowning detection |
CN107330378A (en) * | 2017-06-09 | 2017-11-07 | 湖北天业云商网络科技有限公司 | A kind of driving behavior detecting system based on embedded image processing |
CN107578045A (en) * | 2017-09-19 | 2018-01-12 | 北京工业大学 | A kind of Underwater targets recognition based on machine vision |
CN107729874A (en) * | 2017-11-06 | 2018-02-23 | 青岛博晶微电子科技有限公司 | A kind of swimming pool based on space-time neighborhood characteristics is drowned detection method |
CN109584509A (en) * | 2018-12-27 | 2019-04-05 | 太仓市小车东汽车服务有限公司 | A kind of swimming pool drowning monitoring method combined based on infrared ray with visible light |
TWI657379B (en) * | 2018-05-31 | 2019-04-21 | 國立勤益科技大學 | Intelligent deep learning drowning rescue system |
CN109815805A (en) * | 2018-12-18 | 2019-05-28 | 深圳壹账通智能科技有限公司 | Automatic identification drowned method, apparatus, storage medium and electronic equipment |
CN110084223A (en) * | 2019-05-10 | 2019-08-02 | 南京硅基智能科技有限公司 | A kind of monitoring method and system for swimming pool |
CN110119718A (en) * | 2019-05-15 | 2019-08-13 | 燕山大学 | A kind of overboard detection and Survivable Control System based on deep learning |
CN110155274A (en) * | 2019-05-11 | 2019-08-23 | 北京海豚光波科技开发有限公司 | Intelligent lifesaving system under a kind of swimming-pool water |
CN110363153A (en) * | 2019-07-16 | 2019-10-22 | 广州图普网络科技有限公司 | It paddles detection method, device, server and computer readable storage medium |
WO2019202587A1 (en) * | 2018-04-16 | 2019-10-24 | Lynxight Ltd. | A method and apparatus for swimmer tracking |
WO2020031170A1 (en) * | 2018-08-07 | 2020-02-13 | Lynxight Ltd. | Drowning detection enhanced by swimmer analytics |
CN110793539A (en) * | 2019-11-12 | 2020-02-14 | 浙江大学城市学院 | Drowning detection method and system based on motion component threshold and machine learning |
CN111191486A (en) * | 2018-11-14 | 2020-05-22 | 杭州海康威视数字技术股份有限公司 | Drowning behavior recognition method, monitoring camera and monitoring system |
CN111243237A (en) * | 2020-01-16 | 2020-06-05 | 珠海格力电器股份有限公司 | Drowning monitoring method, equipment, device and storage medium |
CN111339945A (en) * | 2020-02-26 | 2020-06-26 | 贵州安防工程技术研究中心有限公司 | Video-based people group and scatter inspection method and system |
CN112489371A (en) * | 2020-11-26 | 2021-03-12 | 上海天健体育科技发展有限公司 | Swimming pool drowning prevention early warning system based on computer vision |
CN113361364A (en) * | 2021-05-31 | 2021-09-07 | 北京市商汤科技开发有限公司 | Target behavior detection method, device, equipment and storage medium |
CN113468945A (en) * | 2021-03-26 | 2021-10-01 | 厦门大学 | Swimmer drowning detection method |
CN113688724A (en) * | 2021-08-24 | 2021-11-23 | 桂林电子科技大学 | Swimming pool drowning monitoring method based on binocular vision |
CN113838266A (en) * | 2021-09-23 | 2021-12-24 | 广东中星电子有限公司 | Drowning alarm method and device, electronic equipment and computer readable medium |
USD939980S1 (en) | 2019-06-17 | 2022-01-04 | Guard, Inc. | Data and sensor system hub |
CN113901965A (en) * | 2021-12-07 | 2022-01-07 | 广东省科学院智能制造研究所 | Liquid state identification method in liquid separation and liquid separation system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101719216A (en) * | 2009-12-21 | 2010-06-02 | 西安电子科技大学 | Movement human abnormal behavior identification method based on template matching |
CN102542301A (en) * | 2011-12-28 | 2012-07-04 | 浙江大学 | Early-stage drowning behavior detection method based on videos |
CN102722715A (en) * | 2012-05-21 | 2012-10-10 | 华南理工大学 | Tumble detection method based on human body posture state judgment |
-
2013
- 2013-05-17 CN CN2013101890951A patent/CN103413114A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101719216A (en) * | 2009-12-21 | 2010-06-02 | 西安电子科技大学 | Movement human abnormal behavior identification method based on template matching |
CN102542301A (en) * | 2011-12-28 | 2012-07-04 | 浙江大学 | Early-stage drowning behavior detection method based on videos |
CN102722715A (en) * | 2012-05-21 | 2012-10-10 | 华南理工大学 | Tumble detection method based on human body posture state judgment |
Cited By (40)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104902229A (en) * | 2015-05-19 | 2015-09-09 | 吴晗 | Video monitoring method, system and camera shooting monitoring system |
CN105187764A (en) * | 2015-06-26 | 2015-12-23 | 哈尔滨工程大学 | Underwater measurement lifesaving device based on visual analysis |
CN106570854A (en) * | 2015-10-09 | 2017-04-19 | 富士通株式会社 | Water state detection device and method and image processing device |
WO2017130187A1 (en) * | 2016-01-26 | 2017-08-03 | Coral Detection Systems Ltd. | Methods and systems for drowning detection |
US11216654B2 (en) | 2016-01-26 | 2022-01-04 | Coral Detection Systems Ltd. | Methods and systems for drowning detection |
CN106022230B (en) * | 2016-05-11 | 2019-05-14 | 太原理工大学 | A kind of swimming pool drowning event detecting method based on video |
CN106022230A (en) * | 2016-05-11 | 2016-10-12 | 太原理工大学 | Video-based detection method for drowning event in swimming pool |
CN107330378A (en) * | 2017-06-09 | 2017-11-07 | 湖北天业云商网络科技有限公司 | A kind of driving behavior detecting system based on embedded image processing |
CN107578045A (en) * | 2017-09-19 | 2018-01-12 | 北京工业大学 | A kind of Underwater targets recognition based on machine vision |
CN107729874A (en) * | 2017-11-06 | 2018-02-23 | 青岛博晶微电子科技有限公司 | A kind of swimming pool based on space-time neighborhood characteristics is drowned detection method |
CN107729874B (en) * | 2017-11-06 | 2020-10-23 | 青岛联合创智科技有限公司 | Swimming pool drowning detection method based on space-time neighborhood characteristics |
WO2019202587A1 (en) * | 2018-04-16 | 2019-10-24 | Lynxight Ltd. | A method and apparatus for swimmer tracking |
TWI657379B (en) * | 2018-05-31 | 2019-04-21 | 國立勤益科技大學 | Intelligent deep learning drowning rescue system |
WO2020031170A1 (en) * | 2018-08-07 | 2020-02-13 | Lynxight Ltd. | Drowning detection enhanced by swimmer analytics |
US11159769B2 (en) | 2018-08-07 | 2021-10-26 | Lynxight Ltd | Drowning detection enhanced by swimmer analytics |
CN111191486A (en) * | 2018-11-14 | 2020-05-22 | 杭州海康威视数字技术股份有限公司 | Drowning behavior recognition method, monitoring camera and monitoring system |
CN111191486B (en) * | 2018-11-14 | 2023-09-05 | 杭州海康威视数字技术股份有限公司 | Drowning behavior recognition method, monitoring camera and monitoring system |
CN109815805A (en) * | 2018-12-18 | 2019-05-28 | 深圳壹账通智能科技有限公司 | Automatic identification drowned method, apparatus, storage medium and electronic equipment |
CN109584509A (en) * | 2018-12-27 | 2019-04-05 | 太仓市小车东汽车服务有限公司 | A kind of swimming pool drowning monitoring method combined based on infrared ray with visible light |
CN109584509B (en) * | 2018-12-27 | 2020-08-11 | 太仓市小车东汽车服务有限公司 | Swimming pool drowning monitoring method based on combination of infrared rays and visible light |
CN110084223A (en) * | 2019-05-10 | 2019-08-02 | 南京硅基智能科技有限公司 | A kind of monitoring method and system for swimming pool |
CN110155274A (en) * | 2019-05-11 | 2019-08-23 | 北京海豚光波科技开发有限公司 | Intelligent lifesaving system under a kind of swimming-pool water |
CN110119718A (en) * | 2019-05-15 | 2019-08-13 | 燕山大学 | A kind of overboard detection and Survivable Control System based on deep learning |
USD939980S1 (en) | 2019-06-17 | 2022-01-04 | Guard, Inc. | Data and sensor system hub |
USD957966S1 (en) | 2019-06-17 | 2022-07-19 | Guard, Inc. | Tile sensor unit |
CN110363153A (en) * | 2019-07-16 | 2019-10-22 | 广州图普网络科技有限公司 | It paddles detection method, device, server and computer readable storage medium |
CN110363153B (en) * | 2019-07-16 | 2022-08-02 | 广州图普网络科技有限公司 | Wading detection method and device, server and computer readable storage medium |
CN110793539A (en) * | 2019-11-12 | 2020-02-14 | 浙江大学城市学院 | Drowning detection method and system based on motion component threshold and machine learning |
CN111243237A (en) * | 2020-01-16 | 2020-06-05 | 珠海格力电器股份有限公司 | Drowning monitoring method, equipment, device and storage medium |
CN111339945A (en) * | 2020-02-26 | 2020-06-26 | 贵州安防工程技术研究中心有限公司 | Video-based people group and scatter inspection method and system |
CN112489371A (en) * | 2020-11-26 | 2021-03-12 | 上海天健体育科技发展有限公司 | Swimming pool drowning prevention early warning system based on computer vision |
CN113468945A (en) * | 2021-03-26 | 2021-10-01 | 厦门大学 | Swimmer drowning detection method |
CN113361364A (en) * | 2021-05-31 | 2021-09-07 | 北京市商汤科技开发有限公司 | Target behavior detection method, device, equipment and storage medium |
CN113361364B (en) * | 2021-05-31 | 2022-11-01 | 北京市商汤科技开发有限公司 | Target behavior detection method, device, equipment and storage medium |
CN113688724A (en) * | 2021-08-24 | 2021-11-23 | 桂林电子科技大学 | Swimming pool drowning monitoring method based on binocular vision |
CN113688724B (en) * | 2021-08-24 | 2023-03-24 | 桂林电子科技大学 | Swimming pool drowning monitoring method based on binocular vision |
CN113838266B (en) * | 2021-09-23 | 2023-04-07 | 广东中星电子有限公司 | Drowning alarm method and device, electronic equipment and computer readable medium |
CN113838266A (en) * | 2021-09-23 | 2021-12-24 | 广东中星电子有限公司 | Drowning alarm method and device, electronic equipment and computer readable medium |
CN113901965B (en) * | 2021-12-07 | 2022-05-24 | 广东省科学院智能制造研究所 | Liquid state identification method in liquid separation and liquid separation system |
CN113901965A (en) * | 2021-12-07 | 2022-01-07 | 广东省科学院智能制造研究所 | Liquid state identification method in liquid separation and liquid separation system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103413114A (en) | Near-drowning behavior detection method based on support vector machine | |
CN102542301A (en) | Early-stage drowning behavior detection method based on videos | |
CN107818326B (en) | A kind of ship detection method and system based on scene multidimensional characteristic | |
CN102629384B (en) | Method for detecting abnormal behavior during video monitoring | |
CN107748873B (en) | A kind of multimodal method for tracking target merging background information | |
CN103295016B (en) | Behavior recognition method based on depth and RGB information and multi-scale and multidirectional rank and level characteristics | |
CN111091072A (en) | YOLOv 3-based flame and dense smoke detection method | |
CN103279737B (en) | A kind of behavioral value method of fighting based on space-time interest points | |
CN101470809B (en) | Moving object detection method based on expansion mixed gauss model | |
CN103093198B (en) | A kind of crowd density monitoring method and device | |
CN109033934A (en) | A kind of floating on water surface object detecting method based on YOLOv2 network | |
CN110378288A (en) | A kind of multistage spatiotemporal motion object detection method based on deep learning | |
WO2022067668A1 (en) | Fire detection method and system based on video image target detection, and terminal and storage medium | |
CN107330922A (en) | Video moving object detection method of taking photo by plane based on movable information and provincial characteristics | |
CN102043967A (en) | Effective modeling and identification method of moving object behaviors | |
CN114202646A (en) | Infrared image smoking detection method and system based on deep learning | |
CN102096812A (en) | Teacher blackboard writing action detection method for intelligent teaching recording and playing system | |
CN115240069A (en) | Real-time obstacle detection method in full-fog scene | |
CN115631407A (en) | Underwater transparent biological detection based on event camera and color frame image fusion | |
CN105512600A (en) | License plate identification method based on mutual information and characteristic extraction | |
Wang et al. | Detection of early dangerous state in deep water of indoor swimming pool based on surveillance video | |
CN115115672A (en) | Dynamic vision SLAM method based on target detection and feature point speed constraint | |
CN103578112B (en) | A kind of aerator working state detecting method based on video image characteristic | |
CN103886289B (en) | Direction self-adaptive method and system for identifying on-water bridge targets | |
CN105844671B (en) | A kind of fast background relief method under the conditions of change illumination |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20131127 |